docs: record Qwen3.5-9B DFlash Metal timeout (refs #152, #154)

Merged PR #90: feat: integration test — turboquant compressed model

README.md

@@ -30,3 +30,4 @@ See [issues](https://forge.alexanderwhitestone.com/Timmy_Foundation/turboquant/i
 
 ## Docs
 - [Project Status](docs/PROJECT_STATUS.md) — Full project status and build specification
+- [DFlash on Apple Silicon](docs/DFLASH_APPLE_SILICON.md) — MLX benchmark planner, setup commands, and report workflow

benchmarks/bonsai-tool-calling.md

@@ -1,84 +0,0 @@
-# 1-Bit Model Tool Calling Test Results
-
-**Model:** bonsai-1b
-**Date:** 2026-04-15 21:57:29
-**Test cases:** 11
-
-## Summary
-
-| Result | Count |
-|--------|-------|
-| SKIP | 11 |
-
-**Pass rate: 0%** (0/11)
-
-## Results by Difficulty
-
-| Difficulty | PASS | PARTIAL | FAIL | Other |
-|-----------|------|---------|------|-------|
-| 1/5 | 0 | 0 | 0 | 1 |
-| 2/5 | 0 | 0 | 0 | 3 |
-| 3/5 | 0 | 0 | 0 | 5 |
-| 4/5 | 0 | 0 | 0 | 1 |
-| 5/5 | 0 | 0 | 0 | 1 |
-
-## Detailed Results
-
-### ❓ simple-read-1 (difficulty 1/5)
-- **Category:** simple_read
-- **Expected tool:** `read_file`
-- **Actual tool:** `(dry run)`
-
-### ❓ simple-read-with-limit (difficulty 2/5)
-- **Category:** simple_read
-- **Expected tool:** `read_file`
-- **Actual tool:** `(dry run)`
-
-### ❓ terminal-simple (difficulty 2/5)
-- **Category:** terminal_cmd
-- **Expected tool:** `terminal`
-- **Actual tool:** `(dry run)`
-
-### ❓ terminal-pipe (difficulty 3/5)
-- **Category:** terminal_cmd
-- **Expected tool:** `terminal`
-- **Actual tool:** `(dry run)`
-
-### ❓ web-search-simple (difficulty 2/5)
-- **Category:** web_search
-- **Expected tool:** `web_search`
-- **Actual tool:** `(dry run)`
-
-### ❓ multi-tool-select-read (difficulty 3/5)
-- **Category:** multi_tool_select
-- **Expected tool:** `read_file`
-- **Actual tool:** `(dry run)`
-
-### ❓ multi-tool-select-terminal (difficulty 3/5)
-- **Category:** multi_tool_select
-- **Expected tool:** `terminal`
-- **Actual tool:** `(dry run)`
-
-### ❓ multi-tool-select-search (difficulty 3/5)
-- **Category:** multi_tool_select
-- **Expected tool:** `web_search`
-- **Actual tool:** `(dry run)`
-
-### ❓ write-file-with-content (difficulty 3/5)
-- **Category:** nested_params
-- **Expected tool:** `write_file`
-- **Actual tool:** `(dry run)`
-
-### ❓ patch-edit (difficulty 4/5)
-- **Category:** nested_params
-- **Expected tool:** `patch`
-- **Actual tool:** `(dry run)`
-
-### ❓ multi-step-read-then-write (difficulty 5/5)
-- **Category:** multi_step
-- **Expected tool:** `read_file`
-- **Actual tool:** `(dry run)`
-
-## Viability Verdict
-
-**VERDICT: NOT VIABLE** — 1-bit quantization destroys tool calling capability. Recommend minimum 3-bit quantization for tool-using models.

benchmarks/dflash_apple_silicon.py

@@ -0,0 +1,189 @@
+#!/usr/bin/env python3
+"""Apple Silicon DFlash planning helpers and CLI (issue #152)."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import platform
+import subprocess
+from dataclasses import asdict, dataclass
+from pathlib import Path
+from typing import Iterable, Optional
+
+
+@dataclass(frozen=True)
+class DFlashPair:
+    slug: str
+    base_model: str
+    draft_model: str
+    estimated_total_weights_gb: float
+    minimum_recommended_memory_gb: float
+    draft_sliding_window_size: int = 4096
+
+
+SUPPORTED_PAIRS: tuple[DFlashPair, ...] = (
+    DFlashPair(
+        slug="qwen35-4b",
+        base_model="Qwen/Qwen3.5-4B",
+        draft_model="z-lab/Qwen3.5-4B-DFlash",
+        estimated_total_weights_gb=9.68,
+        minimum_recommended_memory_gb=16.0,
+    ),
+    DFlashPair(
+        slug="qwen35-9b",
+        base_model="Qwen/Qwen3.5-9B",
+        draft_model="z-lab/Qwen3.5-9B-DFlash",
+        estimated_total_weights_gb=19.93,
+        minimum_recommended_memory_gb=28.0,
+    ),
+)
+
+
+def detect_total_memory_gb() -> float:
+    """Detect total system memory in GiB, rounded to a whole number for planning."""
+    system = platform.system()
+    if system == "Darwin":
+        mem_bytes = int(subprocess.check_output(["sysctl", "-n", "hw.memsize"]).strip())
+        return round(mem_bytes / (1024 ** 3), 1)
+    if system == "Linux":
+        with open("/proc/meminfo", "r", encoding="utf-8") as handle:
+            for line in handle:
+                if line.startswith("MemTotal:"):
+                    mem_kb = int(line.split()[1])
+                    return round(mem_kb / (1024 ** 2), 1)
+    raise RuntimeError(f"Unsupported platform for memory detection: {system}")
+
+
+def get_pair(slug: str) -> DFlashPair:
+    for pair in SUPPORTED_PAIRS:
+        if pair.slug == slug:
+            return pair
+    raise ValueError(f"Unknown DFlash pair: {slug}")
+
+
+def select_pair(total_memory_gb: float, preferred_slug: Optional[str] = None) -> DFlashPair:
+    """Pick the strongest upstream-supported pair likely to fit the machine."""
+    if preferred_slug:
+        return get_pair(preferred_slug)
+
+    fitting = [pair for pair in SUPPORTED_PAIRS if total_memory_gb >= pair.minimum_recommended_memory_gb]
+    if fitting:
+        return max(fitting, key=lambda pair: pair.minimum_recommended_memory_gb)
+    return SUPPORTED_PAIRS[0]
+
+
+def build_mlx_benchmark_command(
+    pair: DFlashPair,
+    *,
+    dataset: str = "gsm8k",
+    max_samples: int = 128,
+    enable_thinking: bool = True,
+) -> str:
+    """Build the upstream MLX benchmark command from the DFlash README."""
+    parts = [
+        "python -m dflash.benchmark --backend mlx",
+        f"--model {pair.base_model}",
+        f"--draft-model {pair.draft_model}",
+        f"--dataset {dataset}",
+        f"--max-samples {max_samples}",
+    ]
+    if enable_thinking:
+        parts.append("--enable-thinking")
+    parts.append(f"--draft-sliding-window-size {pair.draft_sliding_window_size}")
+    return " \\\n    ".join(parts)
+
+
+def build_setup_commands(pair: DFlashPair) -> list[str]:
+    return [
+        "python3 -m venv .venv-dflash",
+        "source .venv-dflash/bin/activate",
+        "git clone https://github.com/z-lab/dflash.git",
+        "cd dflash",
+        "pip install -e .[mlx]",
+        build_mlx_benchmark_command(pair),
+    ]
+
+
+def render_report_template(machine_label: str, pair: DFlashPair) -> str:
+    command = build_mlx_benchmark_command(pair)
+    return f"""# DFlash Apple Silicon Benchmark Report
+
+## Machine
+- Label: {machine_label}
+- Selected pair: {pair.slug}
+- Base model: {pair.base_model}
+- Draft model: {pair.draft_model}
+- Estimated total weight footprint: {pair.estimated_total_weights_gb:.2f} GB
+
+## Setup
+```bash
+python3 -m venv .venv-dflash
+source .venv-dflash/bin/activate
+git clone https://github.com/z-lab/dflash.git
+cd dflash
+pip install -e .[mlx]
+{command}
+```
+
+## Baseline comparison
+Compare against **plain MLX or llama.cpp speculative decoding** on the same prompt set.
+
+## Results
+- Throughput (tok/s):
+- Peak memory (GB):
+- Notes on acceptance / behavior:
+
+## Verdict
+Worth operationalizing locally?
+- [ ] Yes
+- [ ] No
+- [ ] Needs more data
+
+## Recommendation
+Explain whether this should become part of the local inference stack.
+"""
+
+
+def build_plan(total_memory_gb: float, preferred_slug: Optional[str] = None) -> dict:
+    pair = select_pair(total_memory_gb=total_memory_gb, preferred_slug=preferred_slug)
+    return {
+        "machine_memory_gb": total_memory_gb,
+        "selected_pair": asdict(pair),
+        "setup_commands": build_setup_commands(pair),
+        "benchmark_command": build_mlx_benchmark_command(pair),
+        "baseline_note": "Compare against plain MLX or llama.cpp speculative decoding on the same prompt set.",
+    }
+
+
+def write_output(path: Path, content: str) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    path.write_text(content, encoding="utf-8")
+
+
+def main(argv: Optional[Iterable[str]] = None) -> int:
+    parser = argparse.ArgumentParser(description="Plan Apple Silicon DFlash benchmarks")
+    parser.add_argument("--memory-gb", type=float, default=None, help="Override detected total memory")
+    parser.add_argument("--pair", choices=[pair.slug for pair in SUPPORTED_PAIRS], default=None)
+    parser.add_argument("--machine-label", default="Apple Silicon Mac")
+    parser.add_argument("--format", choices=["json", "markdown"], default="markdown")
+    parser.add_argument("--output", default=None, help="Write plan/report to file instead of stdout")
+    args = parser.parse_args(list(argv) if argv is not None else None)
+
+    memory_gb = args.memory_gb if args.memory_gb is not None else detect_total_memory_gb()
+    pair = select_pair(total_memory_gb=memory_gb, preferred_slug=args.pair)
+
+    if args.format == "json":
+        content = json.dumps(build_plan(memory_gb, preferred_slug=pair.slug), indent=2)
+    else:
+        content = render_report_template(args.machine_label, pair)
+
+    if args.output:
+        write_output(Path(args.output), content)
+    else:
+        print(content)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

benchmarks/reports/dflash_m3max_36gb.md

@@ -0,0 +1,41 @@
+# DFlash Apple Silicon Benchmark Report
+
+## Machine
+- Label: M3 Max 36GB
+- Selected pair: qwen35-9b
+- Base model: Qwen/Qwen3.5-9B
+- Draft model: z-lab/Qwen3.5-9B-DFlash
+- Estimated total weight footprint: 19.93 GB
+
+## Setup
+```bash
+python3 -m venv .venv-dflash
+source .venv-dflash/bin/activate
+git clone https://github.com/z-lab/dflash.git
+cd dflash
+pip install -e .[mlx]
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-9B \
+    --draft-model z-lab/Qwen3.5-9B-DFlash \
+    --dataset gsm8k \
+    --max-samples 128 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## Baseline comparison
+Compare against **plain MLX or llama.cpp speculative decoding** on the same prompt set.
+
+## Results
+- Throughput (tok/s):
+- Peak memory (GB):
+- Notes on acceptance / behavior:
+
+## Verdict
+Worth operationalizing locally?
+- [ ] Yes
+- [ ] No
+- [ ] Needs more data
+
+## Recommendation
+Explain whether this should become part of the local inference stack.

benchmarks/reports/dflash_m3max_36gb_qwen35_4b_pilot.md

@@ -0,0 +1,46 @@
+# DFlash Apple Silicon Pilot — Qwen3.5-4B on M3 Max 36GB
+
+Date: 2026-04-21
+Machine: Apple M3 Max, 36 GB unified memory
+Repo issue: #152
+
+## Command
+
+```bash
+source /tmp/dflash-venv/bin/activate
+cd /tmp/dflash-upstream
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-4B \
+    --draft-model z-lab/Qwen3.5-4B-DFlash \
+    --dataset gsm8k \
+    --max-samples 1 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## Result
+
+- Dataset: `gsm8k`
+- Samples: `1`
+- Baseline throughput: `22.35 tok/s`
+- DFlash throughput: `46.78 tok/s`
+- Decoding speedup: `2.09x`
+- Average acceptance length: `6.48`
+
+Acceptance length histogram:
+
+```text
+['0.3%', '11.1%', '12.7%', '10.4%', '11.7%', '7.6%', '7.0%', '3.8%', '5.1%', '6.3%', '2.8%', '3.8%', '2.2%', '1.9%', '0.9%', '2.5%', '9.8%']
+```
+
+## Caveats
+
+- This is a **pilot**, not a decision-grade benchmark.
+- Only `1` sample was run, so the throughput number is directional.
+- No apples-to-apples baseline against plain MLX or llama.cpp speculative decoding is included yet.
+- The planner still recommends trying `Qwen/Qwen3.5-9B + z-lab/Qwen3.5-9B-DFlash` on this machine for the more meaningful fit test.
+
+## Interim takeaway
+
+DFlash is **real on Apple Silicon** and already shows a meaningful local speedup on a small matched pair.
+A `2.09x` pilot speedup on `Qwen3.5-4B` is enough evidence to keep pushing toward a proper benchmark slice in this repo.

benchmarks/reports/dflash_m3max_36gb_qwen35_9b_timeout.md

@@ -0,0 +1,59 @@
+# DFlash on Apple Silicon Failure Report — Qwen3.5-9B on M3 Max 36GB
+
+Date: 2026-04-21
+Machine: Apple M3 Max, 36 GB unified memory
+Repo issue: #152
+
+## Command
+
+```bash
+source /tmp/dflash-venv/bin/activate
+cd /tmp/dflash-upstream
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-9B \
+    --draft-model z-lab/Qwen3.5-9B-DFlash \
+    --dataset gsm8k \
+    --max-samples 1 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## Outcome
+
+The benchmark did **not** complete successfully on this machine.
+
+### Failure signature
+
+```text
+libc++abi: terminating due to uncaught exception of type std::runtime_error:
+[METAL] Command buffer execution failed:
+Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout)
+```
+
+Additional shutdown noise:
+
+```text
+bash: [11285: 1] tcsetattr: Inappropriate ioctl for device
+resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
+```
+
+## Interpretation
+
+This is strong evidence that the `Qwen/Qwen3.5-9B + z-lab/Qwen3.5-9B-DFlash` pair is **not currently stable** on an M3 Max 36GB Mac under the upstream MLX benchmark path, at least with the default settings used here.
+
+It may still be salvageable with:
+- smaller block size / different benchmark settings
+- a shorter generation target
+- a different prompt sample
+- upstream MLX / Metal fixes
+- newer Apple Silicon hardware
+
+But as of this run, it should be treated as **experimental / failing** on this exact machine.
+
+## Recommendation
+
+For this Mac, the working local proof path is still:
+- `Qwen/Qwen3.5-4B`
+- `z-lab/Qwen3.5-4B-DFlash`
+
+Use the 4B pair for reproducible local validation while the 9B Metal timeout is investigated separately.

benchmarks/test_bonsai_tool_calling.py

@@ -1,709 +0,0 @@
-#!/usr/bin/env python3
-"""
-1-Bit Model Tool Calling Test Suite (Issue #101).
-
-Tests whether quantized/1-bit models can handle structured tool calling.
-Designed to be run against any OpenAI-compatible endpoint (llama-server, Ollama).
-
-The core question: does 1-bit quantization destroy the precise JSON output
-required for tool calling? This suite measures it empirically.
-
-Usage:
-    # Against local llama-server
-    python3 benchmarks/test_bonsai_tool_calling.py \
-        --url http://localhost:8081/v1/chat/completions \
-        --model bonsai-1b
-
-    # Against Ollama
-    python3 benchmarks/test_bonsai_tool_calling.py \
-        --url http://localhost:11434/api/chat \
-        --model bonsai:latest \
-        --backend ollama
-
-    # Dry run (validate test cases without model)
-    python3 benchmarks/test_bonsai_tool_calling.py --dry-run
-"""
-import argparse
-import json
-import os
-import re
-import sys
-import time
-from dataclasses import dataclass, field, asdict
-from enum import Enum
-from typing import List, Dict, Optional, Tuple
-
-import requests
-
-
-class ToolCallCategory(Enum):
-    """Categories of tool call complexity."""
-    SIMPLE_READ = "simple_read"
-    TERMINAL_CMD = "terminal_cmd"
-    WEB_SEARCH = "web_search"
-    MULTI_STEP = "multi_step"
-    NESTED_PARAMS = "nested_params"
-    ARRAY_PARAMS = "array_params"
-    OPTIONAL_PARAMS = "optional_params"
-    MULTI_TOOL_SELECT = "multi_tool_select"
-
-
-class TestResult(Enum):
-    PASS = "PASS"
-    FAIL = "FAIL"
-    PARTIAL = "PARTIAL"
-    TIMEOUT = "TIMEOUT"
-    ERROR = "ERROR"
-    SKIP = "SKIP"
-
-
-# ── Tool schemas (hermes-compatible) ─────────────────────────
-
-TOOL_SCHEMAS = [
-    {
-        "type": "function",
-        "function": {
-            "name": "read_file",
-            "description": "Read a text file with line numbers.",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "path": {"type": "string", "description": "File path to read"},
-                    "offset": {"type": "integer", "description": "Start line (1-indexed)", "default": 1},
-                    "limit": {"type": "integer", "description": "Max lines to read", "default": 500},
-                },
-                "required": ["path"],
-            },
-        },
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "terminal",
-            "description": "Execute a shell command.",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "command": {"type": "string", "description": "Shell command to execute"},
-                    "timeout": {"type": "integer", "description": "Timeout in seconds", "default": 30},
-                    "workdir": {"type": "string", "description": "Working directory"},
-                },
-                "required": ["command"],
-            },
-        },
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "web_search",
-            "description": "Search the web for information.",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "query": {"type": "string", "description": "Search query"},
-                    "max_results": {"type": "integer", "description": "Max results to return", "default": 5},
-                },
-                "required": ["query"],
-            },
-        },
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "write_file",
-            "description": "Write content to a file, creating directories as needed.",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "path": {"type": "string", "description": "File path to write"},
-                    "content": {"type": "string", "description": "Content to write"},
-                },
-                "required": ["path", "content"],
-            },
-        },
-    },
-    {
-        "type": "function",
-        "function": {
-            "name": "patch",
-            "description": "Apply a targeted find-and-replace edit to a file.",
-            "parameters": {
-                "type": "object",
-                "properties": {
-                    "path": {"type": "string", "description": "File path to edit"},
-                    "old_string": {"type": "string", "description": "Text to find"},
-                    "new_string": {"type": "string", "description": "Replacement text"},
-                    "replace_all": {"type": "boolean", "description": "Replace all occurrences", "default": False},
-                },
-                "required": ["path", "old_string", "new_string"],
-            },
-        },
-    },
-]
-
-
-# ── Test case definitions ────────────────────────────────────
-
-@dataclass
-class ToolCallTestCase:
-    """A single tool calling test case."""
-    id: str
-    category: ToolCallCategory
-    prompt: str
-    tools: List[dict]
-    expected_tool: str
-    expected_params: Dict[str, any]
-    param_validators: Dict[str, callable] = field(default_factory=dict)
-    description: str = ""
-    difficulty: int = 1  # 1-5, higher = harder
-
-
-TEST_CASES = [
-    # ── Level 1: Simple reads ──────────────────────────────
-    ToolCallTestCase(
-        id="simple-read-1",
-        category=ToolCallCategory.SIMPLE_READ,
-        prompt="Read the file at /tmp/test.txt",
-        tools=[TOOL_SCHEMAS[0]],
-        expected_tool="read_file",
-        expected_params={"path": "/tmp/test.txt"},
-        description="Exact path, single required param",
-        difficulty=1,
-    ),
-    ToolCallTestCase(
-        id="simple-read-with-limit",
-        category=ToolCallCategory.SIMPLE_READ,
-        prompt="Read the first 10 lines of /var/log/system.log",
-        tools=[TOOL_SCHEMAS[0]],
-        expected_tool="read_file",
-        expected_params={"path": "/var/log/system.log"},
-        param_validators={"limit": lambda v: isinstance(v, int) and v <= 20},
-        description="Required + optional param",
-        difficulty=2,
-    ),
-
-    # ── Level 2: Terminal commands ─────────────────────────
-    ToolCallTestCase(
-        id="terminal-simple",
-        category=ToolCallCategory.TERMINAL_CMD,
-        prompt="List all files in the current directory",
-        tools=[TOOL_SCHEMAS[1]],
-        expected_tool="terminal",
-        expected_params={},
-        param_validators={
-            "command": lambda v: isinstance(v, str) and any(
-                cmd in v for cmd in ["ls", "dir", "find"]
-            )
-        },
-        description="Generate appropriate shell command",
-        difficulty=2,
-    ),
-    ToolCallTestCase(
-        id="terminal-pipe",
-        category=ToolCallCategory.TERMINAL_CMD,
-        prompt="Count how many Python files are in /tmp recursively",
-        tools=[TOOL_SCHEMAS[1]],
-        expected_tool="terminal",
-        expected_params={},
-        param_validators={
-            "command": lambda v: isinstance(v, str) and (
-                "find" in v or "ls" in v or "python" in v or ".py" in v
-            )
-        },
-        description="Needs piped or recursive command",
-        difficulty=3,
-    ),
-
-    # ── Level 3: Web search ────────────────────────────────
-    ToolCallTestCase(
-        id="web-search-simple",
-        category=ToolCallCategory.WEB_SEARCH,
-        prompt="Search for the current price of Bitcoin",
-        tools=[TOOL_SCHEMAS[2]],
-        expected_tool="web_search",
-        expected_params={"query": "Bitcoin price"},
-        param_validators={
-            "query": lambda v: isinstance(v, str) and len(v) > 3 and "bitcoin" in v.lower()
-        },
-        description="Extract search query from natural language",
-        difficulty=2,
-    ),
-
-    # ── Level 4: Multi-tool selection ──────────────────────
-    ToolCallTestCase(
-        id="multi-tool-select-read",
-        category=ToolCallCategory.MULTI_TOOL_SELECT,
-        prompt="Read the file at /etc/hostname",
-        tools=TOOL_SCHEMAS[:3],  # read_file, terminal, web_search
-        expected_tool="read_file",
-        expected_params={"path": "/etc/hostname"},
-        description="Choose correct tool from 3 options",
-        difficulty=3,
-    ),
-    ToolCallTestCase(
-        id="multi-tool-select-terminal",
-        category=ToolCallCategory.MULTI_TOOL_SELECT,
-        prompt="Check how much disk space is available",
-        tools=TOOL_SCHEMAS[:3],
-        expected_tool="terminal",
-        expected_params={},
-        param_validators={
-            "command": lambda v: isinstance(v, str) and any(
-                cmd in v for cmd in ["df", "du", "disk"]
-            )
-        },
-        description="Choose terminal over read_file for system info",
-        difficulty=3,
-    ),
-    ToolCallTestCase(
-        id="multi-tool-select-search",
-        category=ToolCallCategory.MULTI_TOOL_SELECT,
-        prompt="What is the weather in Tokyo right now?",
-        tools=TOOL_SCHEMAS[:3],
-        expected_tool="web_search",
-        expected_params={},
-        param_validators={
-            "query": lambda v: isinstance(v, str) and "weather" in v.lower() and "tokyo" in v.lower()
-        },
-        description="Choose web_search for real-time info",
-        difficulty=3,
-    ),
-
-    # ── Level 5: Nested/complex params ─────────────────────
-    ToolCallTestCase(
-        id="write-file-with-content",
-        category=ToolCallCategory.NESTED_PARAMS,
-        prompt="Create a file at /tmp/hello.txt with the content 'Hello, World!'",
-        tools=[TOOL_SCHEMAS[3]],
-        expected_tool="write_file",
-        expected_params={"path": "/tmp/hello.txt"},
-        param_validators={
-            "content": lambda v: isinstance(v, str) and "hello" in v.lower()
-        },
-        description="Two required string params",
-        difficulty=3,
-    ),
-    ToolCallTestCase(
-        id="patch-edit",
-        category=ToolCallCategory.NESTED_PARAMS,
-        prompt="In the file /tmp/config.yaml, replace 'debug: false' with 'debug: true'",
-        tools=[TOOL_SCHEMAS[4]],
-        expected_tool="patch",
-        expected_params={"path": "/tmp/config.yaml"},
-        param_validators={
-            "old_string": lambda v: isinstance(v, str) and "debug: false" in v,
-            "new_string": lambda v: isinstance(v, str) and "debug: true" in v,
-        },
-        description="Three required params, find-and-replace",
-        difficulty=4,
-    ),
-
-    # ── Level 6: Multi-step reasoning ──────────────────────
-    ToolCallTestCase(
-        id="multi-step-read-then-write",
-        category=ToolCallCategory.MULTI_STEP,
-        prompt="Read /tmp/source.txt and write its contents to /tmp/backup.txt",
-        tools=[TOOL_SCHEMAS[0], TOOL_SCHEMAS[3]],  # read_file + write_file
-        expected_tool="read_file",  # First step should be reading
-        expected_params={"path": "/tmp/source.txt"},
-        description="Requires planning: read first, then write",
-        difficulty=5,
-    ),
-]
-
-
-# ── Test runner ──────────────────────────────────────────────
-
-@dataclass
-class TestRunResult:
-    """Result of running a single test case."""
-    test_id: str
-    category: str
-    difficulty: int
-    result: str  # TestResult value
-    expected_tool: str
-    actual_tool: str
-    expected_params: dict
-    actual_params: dict
-    param_scores: Dict[str, bool] = field(default_factory=dict)
-    response_text: str = ""
-    latency_s: float = 0.0
-    tokens_per_sec: float = 0.0
-    error: str = ""
-    raw_response: dict = field(default_factory=dict)
-
-
-def call_openai_compatible(
-    messages: list,
-    tools: list,
-    url: str,
-    model: str,
-    timeout: int = 120,
-) -> dict:
-    """Call an OpenAI-compatible chat completions endpoint."""
-    payload = {
-        "model": model,
-        "messages": messages,
-        "tools": tools,
-        "tool_choice": "auto",
-        "max_tokens": 512,
-        "temperature": 0.0,
-    }
-    resp = requests.post(url, json=payload, timeout=timeout)
-    resp.raise_for_status()
-    return resp.json()
-
-
-def call_ollama(
-    messages: list,
-    tools: list,
-    url: str,
-    model: str,
-    timeout: int = 120,
-) -> dict:
-    """Call Ollama /api/chat endpoint."""
-    # Convert OpenAI tool format to Ollama format
-    ollama_tools = []
-    for t in tools:
-        fn = t["function"]
-        ollama_tools.append({
-            "type": "function",
-            "function": {
-                "name": fn["name"],
-                "description": fn["description"],
-                "parameters": fn["parameters"],
-            },
-        })
-
-    resp = requests.post(url, json={
-        "model": model,
-        "messages": messages,
-        "tools": ollama_tools,
-        "stream": False,
-    }, timeout=timeout)
-    resp.raise_for_status()
-    data = resp.json()
-
-    # Normalize to OpenAI format
-    result = {"choices": [{"message": {}}]}
-    msg = data.get("message", {})
-    result["choices"][0]["message"]["content"] = msg.get("content", "")
-    if msg.get("tool_calls"):
-        result["choices"][0]["message"]["tool_calls"] = msg["tool_calls"]
-    return result
-
-
-def validate_tool_call(
-    response: dict,
-    test: ToolCallTestCase,
-) -> Tuple[TestResult, str, dict, Dict[str, bool]]:
-    """
-    Validate a model response against a test case.
-
-    Returns: (result, actual_tool, actual_params, param_scores)
-    """
-    try:
-        choice = response["choices"][0]
-        msg = choice["message"]
-    except (KeyError, IndexError):
-        return TestResult.FAIL, "", {}, {}
-
-    # Check if model called a tool
-    tool_calls = msg.get("tool_calls", [])
-    if not tool_calls:
-        # Model responded with text instead — check if it at least mentioned the tool
-        content = msg.get("content", "")
-        if test.expected_tool in content:
-            return TestResult.PARTIAL, "text_only", {"content": content}, {}
-        return TestResult.FAIL, "none", {}, {}
-
-    tc = tool_calls[0]
-    actual_tool = tc.get("function", {}).get("name", "")
-
-    # Parse arguments
-    try:
-        args_str = tc.get("function", {}).get("arguments", "{}")
-        if isinstance(args_str, str):
-            actual_params = json.loads(args_str)
-        else:
-            actual_params = args_str
-    except json.JSONDecodeError:
-        return TestResult.FAIL, actual_tool, {}, {"json_parse": False}
-
-    # Check tool name
-    if actual_tool != test.expected_tool:
-        return TestResult.FAIL, actual_tool, actual_params, {
-            "tool_match": False
-        }
-
-    # Validate expected params
-    param_scores = {"tool_match": True}
-    all_pass = True
-
-    for key, expected_val in test.expected_params.items():
-        if key in actual_params:
-            if actual_params[key] == expected_val:
-                param_scores[f"param_{key}"] = True
-            else:
-                param_scores[f"param_{key}"] = False
-                all_pass = False
-        else:
-            param_scores[f"param_{key}"] = False
-            all_pass = False
-
-    # Run custom validators
-    for key, validator in test.param_validators.items():
-        if key in actual_params:
-            try:
-                passed = validator(actual_params[key])
-                param_scores[f"validator_{key}"] = bool(passed)
-                if not passed:
-                    all_pass = False
-            except Exception:
-                param_scores[f"validator_{key}"] = False
-                all_pass = False
-        else:
-            param_scores[f"validator_{key}"] = False
-            all_pass = False
-
-    if all_pass and len(test.expected_params) > 0:
-        return TestResult.PASS, actual_tool, actual_params, param_scores
-    elif all_pass:
-        # No expected params to check — validators passed
-        return TestResult.PASS, actual_tool, actual_params, param_scores
-    else:
-        return TestResult.PARTIAL, actual_tool, actual_params, param_scores
-
-
-def run_test(
-    test: ToolCallTestCase,
-    url: str,
-    model: str,
-    backend: str = "openai",
-    timeout: int = 120,
-) -> TestRunResult:
-    """Run a single test case against the model."""
-    messages = [{"role": "user", "content": test.prompt}]
-
-    start = time.time()
-    try:
-        if backend == "ollama":
-            response = call_ollama(messages, test.tools, url, model, timeout)
-        else:
-            response = call_openai_compatible(messages, test.tools, url, model, timeout)
-        elapsed = time.time() - start
-
-        result, actual_tool, actual_params, param_scores = validate_tool_call(response, test)
-
-        # Extract text response
-        try:
-            text = response["choices"][0]["message"].get("content", "")
-        except (KeyError, IndexError):
-            text = ""
-
-        return TestRunResult(
-            test_id=test.id,
-            category=test.category.value,
-            difficulty=test.difficulty,
-            result=result.value,
-            expected_tool=test.expected_tool,
-            actual_tool=actual_tool,
-            expected_params=test.expected_params,
-            actual_params=actual_params,
-            param_scores=param_scores,
-            response_text=text[:200],
-            latency_s=round(elapsed, 3),
-            raw_response=response,
-        )
-
-    except requests.exceptions.Timeout:
-        return TestRunResult(
-            test_id=test.id,
-            category=test.category.value,
-            difficulty=test.difficulty,
-            result=TestResult.TIMEOUT.value,
-            expected_tool=test.expected_tool,
-            actual_tool="",
-            expected_params=test.expected_params,
-            actual_params={},
-            error=f"Timeout after {timeout}s",
-        )
-    except Exception as e:
-        return TestRunResult(
-            test_id=test.id,
-            category=test.category.value,
-            difficulty=test.difficulty,
-            result=TestResult.ERROR.value,
-            expected_tool=test.expected_tool,
-            actual_tool="",
-            expected_params=test.expected_params,
-            actual_params={},
-            error=str(e)[:200],
-        )
-
-
-def run_dry_run() -> List[TestRunResult]:
-    """Validate test cases without a model."""
-    results = []
-    for test in TEST_CASES:
-        results.append(TestRunResult(
-            test_id=test.id,
-            category=test.category.value,
-            difficulty=test.difficulty,
-            result=TestResult.SKIP.value,
-            expected_tool=test.expected_tool,
-            actual_tool="(dry run)",
-            expected_params=test.expected_params,
-            actual_params={},
-        ))
-    return results
-
-
-def generate_report(results: List[TestRunResult], model: str) -> str:
-    """Generate markdown report."""
-    lines = [
-        f"# 1-Bit Model Tool Calling Test Results",
-        f"",
-        f"**Model:** {model}",
-        f"**Date:** {time.strftime('%Y-%m-%d %H:%M:%S')}",
-        f"**Test cases:** {len(results)}",
-        f"",
-    ]
-
-    # Summary table
-    by_result = {}
-    for r in results:
-        by_result[r.result] = by_result.get(r.result, 0) + 1
-
-    lines.append("## Summary")
-    lines.append("")
-    lines.append("| Result | Count |")
-    lines.append("|--------|-------|")
-    for result, count in sorted(by_result.items()):
-        lines.append(f"| {result} | {count} |")
-    lines.append("")
-
-    pass_count = by_result.get("PASS", 0)
-    total = len(results)
-    pass_rate = (pass_count / total * 100) if total > 0 else 0
-    lines.append(f"**Pass rate: {pass_rate:.0f}%** ({pass_count}/{total})")
-    lines.append("")
-
-    # By difficulty
-    lines.append("## Results by Difficulty")
-    lines.append("")
-    lines.append("| Difficulty | PASS | PARTIAL | FAIL | Other |")
-    lines.append("|-----------|------|---------|------|-------|")
-    for diff in range(1, 6):
-        diff_results = [r for r in results if r.difficulty == diff]
-        if not diff_results:
-            continue
-        p = sum(1 for r in diff_results if r.result == "PASS")
-        pa = sum(1 for r in diff_results if r.result == "PARTIAL")
-        f = sum(1 for r in diff_results if r.result in ("FAIL", "ERROR", "TIMEOUT"))
-        o = len(diff_results) - p - pa - f
-        lines.append(f"| {diff}/5 | {p} | {pa} | {f} | {o} |")
-    lines.append("")
-
-    # Detailed results
-    lines.append("## Detailed Results")
-    lines.append("")
-    for r in results:
-        icon = {"PASS": "✅", "PARTIAL": "⚠️", "FAIL": "❌", "ERROR": "💥", "TIMEOUT": "⏱"}.get(r.result, "❓")
-        lines.append(f"### {icon} {r.test_id} (difficulty {r.difficulty}/5)")
-        lines.append(f"- **Category:** {r.category}")
-        lines.append(f"- **Expected tool:** `{r.expected_tool}`")
-        lines.append(f"- **Actual tool:** `{r.actual_tool}`")
-        if r.latency_s > 0:
-            lines.append(f"- **Latency:** {r.latency_s}s")
-        if r.param_scores:
-            lines.append(f"- **Param scores:** {json.dumps(r.param_scores)}")
-        if r.error:
-            lines.append(f"- **Error:** {r.error}")
-        lines.append("")
-
-    # Viability verdict
-    lines.append("## Viability Verdict")
-    lines.append("")
-    if pass_rate >= 80:
-        lines.append("**VERDICT: VIABLE** — 1-bit model can handle tool calling for production use.")
-    elif pass_rate >= 50:
-        lines.append("**VERDICT: CONDITIONALLY VIABLE** — Works for simple tools, struggles with complex params. Consider for edge deployment with guardrails.")
-    elif pass_rate >= 20:
-        lines.append("**VERDICT: MARGINAL** — Can select correct tool sometimes, but parameter accuracy is too low for production. Investigate alternative quantization (2-bit, 3-bit).")
-    else:
-        lines.append("**VERDICT: NOT VIABLE** — 1-bit quantization destroys tool calling capability. Recommend minimum 3-bit quantization for tool-using models.")
-    lines.append("")
-
-    return "\n".join(lines)
-
-
-def main():
-    parser = argparse.ArgumentParser(description="Test tool calling on 1-bit models")
-    parser.add_argument("--url", default="http://localhost:8081/v1/chat/completions",
-                        help="Model API endpoint")
-    parser.add_argument("--model", default="bonsai-1b", help="Model name")
-    parser.add_argument("--backend", default="openai", choices=["openai", "ollama"],
-                        help="API backend type")
-    parser.add_argument("--timeout", type=int, default=120, help="Request timeout in seconds")
-    parser.add_argument("--dry-run", action="store_true", help="Validate tests without model")
-    parser.add_argument("--output", default="benchmarks/bonsai-tool-calling-results.json",
-                        help="Output file for results")
-    parser.add_argument("--report", default="benchmarks/bonsai-tool-calling.md",
-                        help="Output file for markdown report")
-    parser.add_argument("--test-id", help="Run a single test by ID")
-
-    args = parser.parse_args()
-
-    print("=" * 60)
-    print("  1-Bit Model Tool Calling Test Suite")
-    print("=" * 60)
-
-    if args.dry_run:
-        print("\n[DRY RUN] Validating test cases...")
-        results = run_dry_run()
-        print(f"  {len(results)} test cases validated")
-        for r in results:
-            print(f"  ✓ {r.test_id} — expects {r.expected_tool} (difficulty {r.difficulty}/5)")
-    else:
-        print(f"\nModel: {args.model}")
-        print(f"Endpoint: {args.url}")
-        print(f"Backend: {args.backend}")
-        print()
-
-        tests = TEST_CASES
-        if args.test_id:
-            tests = [t for t in tests if t.id == args.test_id]
-            if not tests:
-                print(f"Test '{args.test_id}' not found")
-                sys.exit(1)
-
-        results = []
-        for i, test in enumerate(tests):
-            print(f"  [{i+1}/{len(tests)}] {test.id} (difficulty {test.difficulty}/5)... ", end="", flush=True)
-            result = run_test(test, args.url, args.model, args.backend, args.timeout)
-            results.append(result)
-            icon = {"PASS": "✅", "PARTIAL": "⚠️", "FAIL": "❌", "ERROR": "💥", "TIMEOUT": "⏱"}.get(result.result, "❓")
-            print(f"{icon} {result.result} ({result.latency_s}s)")
-
-    # Save results
-    os.makedirs(os.path.dirname(args.output) or ".", exist_ok=True)
-    with open(args.output, "w") as f:
-        json.dump([asdict(r) for r in results], f, indent=2)
-    print(f"\nResults saved to {args.output}")
-
-    # Generate report
-    report = generate_report(results, args.model)
-    with open(args.report, "w") as f:
-        f.write(report)
-    print(f"Report saved to {args.report}")
-
-    # Print summary
-    pass_count = sum(1 for r in results if r.result == "PASS")
-    total = len(results)
-    print(f"\n{'='*60}")
-    print(f"  Results: {pass_count}/{total} passed ({pass_count/total*100:.0f}%)")
-
-
-if __name__ == "__main__":
-    main()

docs/DFLASH_APPLE_SILICON.md

@@ -0,0 +1,125 @@
+# DFlash on Apple Silicon
+
+This repo now carries a **Gitea-first benchmark harness** for evaluating whether upstream **DFlash on MLX** is worth adding to the local Apple Silicon inference stack.
+
+## Why
+
+The headline `Kimi K2.6 + DFlash` benchmark was measured on `8x MI300X` with huge RAM and ROCm patches. That exact recipe is not a fit for a `36 GB` Apple Silicon Mac.
+
+What *is* relevant locally is the upstream `z-lab/dflash` MLX path, which can benchmark smaller matched target/draft pairs that fit on Apple Silicon.
+
+## Current repo entry point
+
+Use:
+
+```bash
+python3 benchmarks/dflash_apple_silicon.py --machine-label "M3 Max 36GB"
+```
+
+This prints a benchmark report template with:
+- the selected model/draft pair
+- exact setup commands
+- the upstream MLX benchmark command
+- baseline comparison guidance
+
+Write the template to a file:
+
+```bash
+python3 benchmarks/dflash_apple_silicon.py \
+  --machine-label "M3 Max 36GB" \
+  --output benchmarks/reports/dflash_m3max_36gb.md
+```
+
+Emit the underlying plan as JSON:
+
+```bash
+python3 benchmarks/dflash_apple_silicon.py --format json
+```
+
+## Selection logic
+
+Today the planner uses two upstream-supported MLX pairs:
+
+- `qwen35-9b`
+  - base: `Qwen/Qwen3.5-9B`
+  - draft: `z-lab/Qwen3.5-9B-DFlash`
+  - chosen for ~28 GB+ machines
+- `qwen35-4b`
+  - base: `Qwen/Qwen3.5-4B`
+  - draft: `z-lab/Qwen3.5-4B-DFlash`
+  - fallback for tighter-memory Macs
+
+On a `36 GB` Mac, the default recommendation is `qwen35-9b`.
+
+## Pilot result already landed
+
+A first live Apple Silicon run has already been captured in:
+
+- `benchmarks/reports/dflash_m3max_36gb_qwen35_4b_pilot.md`
+
+Pilot command:
+
+```bash
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-4B \
+    --draft-model z-lab/Qwen3.5-4B-DFlash \
+    --dataset gsm8k \
+    --max-samples 1 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+Pilot outcome on this Mac:
+
+- baseline throughput: `22.35 tok/s`
+- DFlash throughput: `46.78 tok/s`
+- decoding speedup: `2.09x`
+
+Treat that as a **directional proof**, not a final decision benchmark. The next step is the fuller comparison slice against plain MLX or llama.cpp speculative decoding.
+
+## Known 9B failure on this machine
+
+A follow-up live run with:
+
+- `Qwen/Qwen3.5-9B`
+- `z-lab/Qwen3.5-9B-DFlash`
+
+failed on this same M3 Max 36GB Mac with:
+
+```text
+[METAL] Command buffer execution failed:
+Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout)
+```
+
+That failure is recorded in:
+
+- `benchmarks/reports/dflash_m3max_36gb_qwen35_9b_timeout.md`
+
+So the current guidance is:
+- treat `qwen35-9b` as **experimental** on this machine
+- treat `qwen35-4b` as the current **known-working local proof path**
+- keep the issue open until we either stabilize the 9B path or clearly rule it out for this hardware tier
+
+## Upstream benchmark command
+
+The harness uses the upstream MLX benchmark syntax from `z-lab/dflash`:
+
+```bash
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-9B \
+    --draft-model z-lab/Qwen3.5-9B-DFlash \
+    --dataset gsm8k \
+    --max-samples 128 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## What remains
+
+This PR adds the **planner + report template** so the benchmark is reproducible from the repo.
+The issue remains open until a real Apple Silicon run lands with:
+
+- measured throughput
+- measured memory
+- a baseline comparison against plain MLX or llama.cpp speculative decoding
+- a recommendation on whether to operationalize DFlash locally

evolution/quant_selector.py

@@ -0,0 +1,548 @@
+"""Auto-select TurboQuant compression level based on available VRAM/RAM.
+
+Detects hardware resources at startup and picks the highest quality
+quantization level that fits within available memory. Supports Apple
+Silicon unified memory, NVIDIA GPUs (via nvidia-smi), and CPU-only fallback.
+
+Usage:
+    from evolution.quant_selector import select_quant_level
+
+    selection = select_quant_level(model_size_gb=14.0, context_length=32768)
+    print(selection.level)       # "turbo4"
+    print(selection.reasoning)   # "M4 Max 36GB unified: turbo4 fits 14.0GB model + ..."
+    print(selection.env_vars)    # {"TURBO_LAYER_ADAPTIVE": "7"}
+"""
+
+import logging
+import os
+import platform
+import subprocess
+import sys
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Optional
+
+logger = logging.getLogger(__name__)
+
+
+# ── Quant Level Definitions ───────────────────────────────────────────────────
+
+@dataclass
+class QuantLevel:
+    """A TurboQuant compression level with its memory characteristics."""
+    name: str               # e.g. "turbo4"
+    bits_per_channel: float # e.g. 3.5 for turbo4
+    compression_ratio: float  # vs uncompressed KV cache
+    quality_label: str      # "best", "high", "balanced", "fast"
+    layer_adaptive: int     # TURBO_LAYER_ADAPTIVE value (0-7)
+    kv_type: str            # -ctk/-ctv flag value
+    min_memory_headroom_gb: float  # Minimum free memory to recommend this level
+    description: str = ""
+
+
+# Ordered from highest quality to most aggressive compression
+QUANT_LEVELS = [
+    QuantLevel(
+        name="turbo4",
+        bits_per_channel=3.5,
+        compression_ratio=4.2,
+        quality_label="best",
+        layer_adaptive=7,
+        kv_type="turbo4",
+        min_memory_headroom_gb=4.0,
+        description="PolarQuant + QJL 4-bit. Best quality, ~4.2x KV compression."
+    ),
+    QuantLevel(
+        name="turbo3",
+        bits_per_channel=2.5,
+        compression_ratio=6.0,
+        quality_label="high",
+        layer_adaptive=5,
+        kv_type="turbo3",
+        min_memory_headroom_gb=3.0,
+        description="3-bit TurboQuant. High quality, ~6x KV compression."
+    ),
+    QuantLevel(
+        name="turbo2",
+        bits_per_channel=1.5,
+        compression_ratio=10.0,
+        quality_label="balanced",
+        layer_adaptive=3,
+        kv_type="turbo2",
+        min_memory_headroom_gb=2.0,
+        description="2-bit TurboQuant. Balanced, ~10x KV compression."
+    ),
+    QuantLevel(
+        name="q4_0",
+        bits_per_channel=4.0,
+        compression_ratio=3.5,
+        quality_label="fast",
+        layer_adaptive=0,
+        kv_type="q4_0",
+        min_memory_headroom_gb=1.5,
+        description="Standard 4-bit quant. Fast fallback, no TurboQuant."
+    ),
+]
+
+
+# ── Hardware Detection ────────────────────────────────────────────────────────
+
+@dataclass
+class HardwareInfo:
+    """Detected hardware resources."""
+    total_memory_gb: float
+    available_memory_gb: float
+    gpu_memory_gb: Optional[float] = None
+    gpu_name: Optional[str] = None
+    is_apple_silicon: bool = False
+    chip_name: Optional[str] = None
+    cpu_cores: int = 0
+    detection_method: str = ""
+
+
+def detect_hardware() -> HardwareInfo:
+    """Detect available memory and GPU resources."""
+    system = platform.system()
+
+    if system == "Darwin":
+        return _detect_apple_silicon()
+    elif system == "Linux":
+        return _detect_linux()
+    else:
+        return _detect_generic(system)
+
+
+def _detect_apple_silicon() -> HardwareInfo:
+    """Detect Apple Silicon unified memory."""
+    info = HardwareInfo(
+        total_memory_gb=0,
+        available_memory_gb=0,
+        is_apple_silicon=True,
+        detection_method="sysctl",
+    )
+
+    try:
+        # Get total memory
+        result = subprocess.run(
+            ["sysctl", "-n", "hw.memsize"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            info.total_memory_gb = int(result.stdout.strip()) / (1024**3)
+
+        # Get chip name
+        result = subprocess.run(
+            ["sysctl", "-n", "machdep.cpu.brand_string"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            info.chip_name = result.stdout.strip()
+
+        # Try to get GPU name (Apple Silicon)
+        result = subprocess.run(
+            ["system_profiler", "SPDisplaysDataType"],
+            capture_output=True, text=True, timeout=10
+        )
+        if result.returncode == 0:
+            for line in result.stdout.split("\n"):
+                if "Chipset" in line or "GPU" in line:
+                    info.gpu_name = line.split(":")[-1].strip()
+                    break
+
+        # Estimate available memory (vm_stat)
+        result = subprocess.run(
+            ["vm_stat"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            page_size = 4096  # macOS default
+            free_pages = 0
+            for line in result.stdout.split("\n"):
+                if "Pages free:" in line:
+                    try:
+                        free_pages = int(line.split(":")[-1].strip().rstrip("."))
+                    except ValueError:
+                        pass
+            # Available ≈ free + some speculative (conservative: just free)
+            info.available_memory_gb = (free_pages * page_size) / (1024**3)
+
+        # Fallback if vm_stat parsing failed
+        if info.available_memory_gb < 1:
+            # Conservative: 70% of total
+            info.available_memory_gb = info.total_memory_gb * 0.70
+
+        # Apple Silicon shares memory — GPU memory = total memory
+        info.gpu_memory_gb = info.total_memory_gb
+
+        # Detect CPU cores
+        result = subprocess.run(
+            ["sysctl", "-n", "hw.ncpu"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            info.cpu_cores = int(result.stdout.strip())
+
+    except Exception as e:
+        logger.warning(f"Apple Silicon detection failed: {e}")
+        # Fallback
+        info.total_memory_gb = 16.0
+        info.available_memory_gb = 12.0
+        info.detection_method = "fallback"
+
+    return info
+
+
+def _detect_linux() -> HardwareInfo:
+    """Detect Linux system with optional NVIDIA GPU."""
+    info = HardwareInfo(
+        total_memory_gb=0,
+        available_memory_gb=0,
+        detection_method="proc",
+    )
+
+    try:
+        # Read /proc/meminfo
+        with open("/proc/meminfo", "r") as f:
+            meminfo = f.read()
+
+        for line in meminfo.split("\n"):
+            if line.startswith("MemTotal:"):
+                kb = int(line.split()[1])
+                info.total_memory_gb = kb / (1024 * 1024)
+            elif line.startswith("MemAvailable:"):
+                kb = int(line.split()[1])
+                info.available_memory_gb = kb / (1024 * 1024)
+
+        # CPU cores
+        info.cpu_cores = os.cpu_count() or 1
+
+        # Check for NVIDIA GPU
+        try:
+            result = subprocess.run(
+                ["nvidia-smi", "--query-gpu=name,memory.total,memory.free",
+                 "--format=csv,noheader,nounits"],
+                capture_output=True, text=True, timeout=10
+            )
+            if result.returncode == 0 and result.stdout.strip():
+                lines = result.stdout.strip().split("\n")
+                if lines:
+                    parts = lines[0].split(", ")
+                    if len(parts) >= 3:
+                        info.gpu_name = parts[0].strip()
+                        info.gpu_memory_gb = float(parts[1]) / 1024  # MB to GB
+                        gpu_free = float(parts[2]) / 1024
+                        # Use GPU free for VRAM-based selection
+                        info.available_memory_gb = max(info.available_memory_gb, gpu_free)
+                        info.detection_method = "nvidia-smi"
+        except (FileNotFoundError, subprocess.TimeoutExpired):
+            pass  # No NVIDIA GPU
+
+    except Exception as e:
+        logger.warning(f"Linux detection failed: {e}")
+        info.total_memory_gb = 16.0
+        info.available_memory_gb = 12.0
+        info.detection_method = "fallback"
+
+    return info
+
+
+def _detect_generic(system: str) -> HardwareInfo:
+    """Fallback detection for unknown systems."""
+    import psutil
+    mem = psutil.virtual_memory()
+    return HardwareInfo(
+        total_memory_gb=mem.total / (1024**3),
+        available_memory_gb=mem.available / (1024**3),
+        cpu_cores=os.cpu_count() or 1,
+        detection_method="psutil",
+    )
+
+
+# ── KV Cache Memory Estimation ───────────────────────────────────────────────
+
+def estimate_kv_cache_gb(
+    context_length: int,
+    num_layers: int = 48,
+    num_kv_heads: int = 8,
+    head_dim: int = 128,
+    bits_per_channel: float = 3.5,
+) -> float:
+    """Estimate KV cache memory for given parameters.
+
+    Formula: 2 (K+V) × layers × kv_heads × head_dim × context_length × bits/8
+    """
+    bytes_per_element = bits_per_channel / 8.0
+    total_bytes = 2 * num_layers * num_kv_heads * head_dim * context_length * bytes_per_element
+    return total_bytes / (1024**3)
+
+
+def estimate_model_memory_gb(model_size_gb: float, quant_type: str = "q4_k_m") -> float:
+    """Estimate model weights memory. Returns loaded size in GB.
+
+    This is a rough estimate — actual depends on exact quant format.
+    """
+    # Common quant ratios (vs fp16)
+    quant_multipliers = {
+        "f16": 1.0,
+        "q8_0": 0.5,
+        "q6_k": 0.42,
+        "q5_k_m": 0.37,
+        "q4_k_m": 0.32,
+        "q3_k_m": 0.27,
+        "q2_k": 0.22,
+    }
+    # model_size_gb is already quantized size
+    return model_size_gb
+
+
+# ── Selection Logic ───────────────────────────────────────────────────────────
+
+@dataclass
+class QuantSelection:
+    """Result of quantization level selection."""
+    level: QuantLevel
+    hardware: HardwareInfo
+    reasoning: str
+    total_required_gb: float
+    available_gb: float
+    headroom_gb: float
+    env_vars: dict = field(default_factory=dict)
+    server_flags: dict = field(default_factory=dict)
+    warnings: list = field(default_factory=list)
+
+
+def select_quant_level(
+    model_size_gb: float = 14.0,
+    context_length: int = 32768,
+    num_layers: int = 48,
+    num_kv_heads: int = 8,
+    head_dim: int = 128,
+    preferred_level: Optional[str] = None,
+    force_cpu: bool = False,
+) -> QuantSelection:
+    """Select the best quantization level for available hardware.
+
+    Args:
+        model_size_gb: Size of the model weights in GB
+        context_length: Target context length
+        num_layers: Number of transformer layers
+        num_kv_heads: Number of KV attention heads
+        head_dim: Dimension per attention head
+        preferred_level: Force a specific level (still checks if it fits)
+        force_cpu: If True, ignore GPU memory
+
+    Returns:
+        QuantSelection with the chosen level and reasoning
+    """
+    hw = detect_hardware()
+
+    if force_cpu:
+        hw.gpu_memory_gb = None
+        hw.gpu_name = None
+
+    # Use the most restrictive memory constraint
+    # For Apple Silicon: unified memory, use total
+    # For NVIDIA: use GPU VRAM
+    # For CPU-only: use system RAM
+    if hw.gpu_memory_gb and hw.gpu_name:
+        memory_pool_gb = hw.gpu_memory_gb
+        memory_label = f"{hw.gpu_name} {hw.gpu_memory_gb:.0f}GB VRAM"
+    elif hw.is_apple_silicon:
+        memory_pool_gb = hw.total_memory_gb
+        memory_label = f"{hw.chip_name or 'Apple Silicon'} {hw.total_memory_gb:.0f}GB unified"
+    else:
+        memory_pool_gb = hw.total_memory_gb
+        memory_label = f"{hw.cpu_cores}c CPU {hw.total_memory_gb:.0f}GB RAM"
+
+    model_mem = estimate_model_memory_gb(model_size_gb)
+
+    # Try levels from best to most compressed
+    chosen = None
+    for level in QUANT_LEVELS:
+        if preferred_level and level.name != preferred_level:
+            continue
+
+        kv_mem = estimate_kv_cache_gb(
+            context_length, num_layers, num_kv_heads, head_dim,
+            level.bits_per_channel
+        )
+        total_required = model_mem + kv_mem
+        headroom = memory_pool_gb - total_required
+
+        if headroom >= level.min_memory_headroom_gb:
+            chosen = level
+            break
+
+        if preferred_level and level.name == preferred_level:
+            # User forced this level but it doesn't fit
+            chosen = level
+            break
+
+    if chosen is None:
+        # Nothing fits — pick the most aggressive compression, not the q4_0 fallback.
+        chosen = max(QUANT_LEVELS, key=lambda level: level.compression_ratio)
+        logger.warning(f"No quant level fits in {memory_pool_gb:.1f}GB. Using {chosen.name}.")
+
+    # Calculate final numbers
+    kv_mem = estimate_kv_cache_gb(
+        context_length, num_layers, num_kv_heads, head_dim,
+        chosen.bits_per_channel
+    )
+    total_required = model_mem + kv_mem
+    headroom = memory_pool_gb - total_required
+
+    # Build reasoning
+    reasoning_parts = [
+        f"{memory_label}:",
+        f"{chosen.name} ({chosen.quality_label}, {chosen.bits_per_channel:.1f}b/ch,",
+        f"{chosen.compression_ratio:.1f}x compression)",
+        f"fits {model_mem:.1f}GB model + {kv_mem:.1f}GB KV cache",
+        f"@ {context_length}K context = {total_required:.1f}GB / {memory_pool_gb:.0f}GB",
+        f"({headroom:.1f}GB headroom)"
+    ]
+    reasoning = " ".join(reasoning_parts)
+
+    # Build environment variables for llama.cpp
+    env_vars = {
+        "TURBO_LAYER_ADAPTIVE": str(chosen.layer_adaptive),
+    }
+
+    # Build server flags
+    server_flags = {
+        "-ctk": chosen.kv_type,
+        "-ctv": chosen.kv_type,
+        "-c": str(context_length),
+    }
+
+    # Warnings
+    warnings = []
+    if headroom < 2.0:
+        warnings.append(
+            f"Low headroom ({headroom:.1f}GB). Consider reducing context length or model size."
+        )
+    if headroom < 0:
+        warnings.append(
+            f"OVERCOMMITTED: needs {total_required:.1f}GB but only {memory_pool_gb:.0f}GB available. "
+            f"Inference may fail or swap heavily."
+        )
+
+    selection = QuantSelection(
+        level=chosen,
+        hardware=hw,
+        reasoning=reasoning,
+        total_required_gb=total_required,
+        available_gb=memory_pool_gb,
+        headroom_gb=headroom,
+        env_vars=env_vars,
+        server_flags=server_flags,
+        warnings=warnings,
+    )
+
+    logger.info(f"Quant selection: {reasoning}")
+    for w in warnings:
+        logger.warning(w)
+
+    return selection
+
+
+# ── CLI ───────────────────────────────────────────────────────────────────────
+
+def main():
+    """CLI entry point for quant level selection."""
+    import argparse
+    import json
+
+    parser = argparse.ArgumentParser(
+        description="Auto-select TurboQuant compression level based on available hardware"
+    )
+    parser.add_argument("--model-size", type=float, default=14.0,
+                        help="Model size in GB (default: 14.0)")
+    parser.add_argument("--context", type=int, default=32768,
+                        help="Target context length (default: 32768)")
+    parser.add_argument("--layers", type=int, default=48,
+                        help="Number of transformer layers (default: 48)")
+    parser.add_argument("--kv-heads", type=int, default=8,
+                        help="Number of KV attention heads (default: 8)")
+    parser.add_argument("--head-dim", type=int, default=128,
+                        help="Dimension per attention head (default: 128)")
+    parser.add_argument("--prefer", type=str, default=None,
+                        choices=[l.name for l in QUANT_LEVELS],
+                        help="Prefer a specific quant level")
+    parser.add_argument("--force-cpu", action="store_true",
+                        help="Ignore GPU, use CPU memory only")
+    parser.add_argument("--json", action="store_true",
+                        help="JSON output for automation")
+    parser.add_argument("--detect-only", action="store_true",
+                        help="Only detect hardware, don't select")
+    args = parser.parse_args()
+
+    logging.basicConfig(level=logging.INFO, format="%(message)s")
+
+    if args.detect_only:
+        hw = detect_hardware()
+        if args.json:
+            print(json.dumps(hw.__dict__, default=str, indent=2))
+        else:
+            print(f"Total memory:    {hw.total_memory_gb:.1f} GB")
+            print(f"Available:       {hw.available_memory_gb:.1f} GB")
+            if hw.gpu_memory_gb:
+                print(f"GPU memory:      {hw.gpu_memory_gb:.1f} GB")
+            if hw.gpu_name:
+                print(f"GPU:             {hw.gpu_name}")
+            if hw.is_apple_silicon:
+                print(f"Chip:            {hw.chip_name or 'Apple Silicon'}")
+            print(f"CPU cores:       {hw.cpu_cores}")
+            print(f"Detection:       {hw.detection_method}")
+        return
+
+    selection = select_quant_level(
+        model_size_gb=args.model_size,
+        context_length=args.context,
+        num_layers=args.layers,
+        num_kv_heads=args.kv_heads,
+        head_dim=args.head_dim,
+        preferred_level=args.prefer,
+        force_cpu=args.force_cpu,
+    )
+
+    if args.json:
+        result = {
+            "level": selection.level.name,
+            "bits_per_channel": selection.level.bits_per_channel,
+            "compression_ratio": selection.level.compression_ratio,
+            "quality": selection.level.quality_label,
+            "reasoning": selection.reasoning,
+            "total_required_gb": round(selection.total_required_gb, 2),
+            "available_gb": round(selection.available_gb, 1),
+            "headroom_gb": round(selection.headroom_gb, 2),
+            "env_vars": selection.env_vars,
+            "server_flags": selection.server_flags,
+            "warnings": selection.warnings,
+            "hardware": {
+                "total_memory_gb": round(selection.hardware.total_memory_gb, 1),
+                "gpu_name": selection.hardware.gpu_name,
+                "is_apple_silicon": selection.hardware.is_apple_silicon,
+                "chip_name": selection.hardware.chip_name,
+                "cpu_cores": selection.hardware.cpu_cores,
+            },
+        }
+        print(json.dumps(result, indent=2))
+    else:
+        print(f"Selected: {selection.level.name} ({selection.level.quality_label})")
+        print(f"  {selection.reasoning}")
+        print()
+        print(f"Environment variables:")
+        for k, v in selection.env_vars.items():
+            print(f"  export {k}={v}")
+        print()
+        print(f"Server flags:")
+        for k, v in selection.server_flags.items():
+            print(f"  {k} {v}")
+        if selection.warnings:
+            print()
+            for w in selection.warnings:
+                print(f"  WARNING: {w}")
+
+
+if __name__ == "__main__":
+    main()

tests/conftest.py

@@ -0,0 +1,3 @@
+"""Pytest configuration for turboquant."""
+import sys, os
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))

tests/test_bonsai_tool_calling.py

@@ -1,236 +0,0 @@
-"""
-Test suite for 1-bit model tool calling validation (issue #101).
-
-Tests the test harness itself — validates test case structure,
-tool schema compatibility, and result generation. The actual
-model inference tests require a running model server.
-
-Usage:
-    pytest tests/test_bonsai_tool_calling.py -v
-    pytest tests/test_bonsai_tool_calling.py -v -k live  # if server available
-"""
-import json
-import os
-import sys
-import unittest
-from unittest.mock import patch, MagicMock
-
-import pytest
-
-# Add benchmarks to path — resolve relative to project root
-_PROJECT_ROOT = os.path.join(os.path.dirname(__file__), "..")
-sys.path.insert(0, os.path.join(_PROJECT_ROOT, "benchmarks"))
-
-# Import with absolute path to avoid collision with this test module
-import importlib.util
-_spec = importlib.util.spec_from_file_location(
-    "bonsai_tool_calling",
-    os.path.join(_PROJECT_ROOT, "benchmarks", "test_bonsai_tool_calling.py"),
-)
-_btc = importlib.util.module_from_spec(_spec)
-_spec.loader.exec_module(_btc)
-
-TOOL_SCHEMAS = _btc.TOOL_SCHEMAS
-TEST_CASES = _btc.TEST_CASES
-ToolCallCategory = _btc.ToolCallCategory
-TestResult = _btc.TestResult
-ToolCallTestCase = _btc.ToolCallTestCase
-TestRunResult = _btc.TestRunResult
-validate_tool_call = _btc.validate_tool_call
-run_dry_run = _btc.run_dry_run
-generate_report = _btc.generate_report
-
-
-class TestToolSchemas(unittest.TestCase):
-    """Validate tool schemas are well-formed."""
-
-    def test_schemas_serialize_to_json(self):
-        serialized = json.dumps(TOOL_SCHEMAS)
-        parsed = json.loads(serialized)
-        assert len(parsed) == len(TOOL_SCHEMAS)
-
-    def test_each_schema_has_required_fields(self):
-        for tool in TOOL_SCHEMAS:
-            assert tool["type"] == "function"
-            fn = tool["function"]
-            assert "name" in fn
-            assert "description" in fn
-            assert "parameters" in fn
-            assert fn["parameters"]["type"] == "object"
-            assert "properties" in fn["parameters"]
-            assert "required" in fn["parameters"]
-
-    def test_tool_names_are_unique(self):
-        names = [t["function"]["name"] for t in TOOL_SCHEMAS]
-        assert len(names) == len(set(names)), f"Duplicate tool names: {names}"
-
-
-class TestTestCaseStructure(unittest.TestCase):
-    """Validate test case definitions."""
-
-    def test_all_categories_covered(self):
-        categories = {tc.category for tc in TEST_CASES}
-        assert ToolCallCategory.SIMPLE_READ in categories
-        assert ToolCallCategory.TERMINAL_CMD in categories
-        assert ToolCallCategory.WEB_SEARCH in categories
-        assert ToolCallCategory.MULTI_TOOL_SELECT in categories
-        assert ToolCallCategory.NESTED_PARAMS in categories
-
-    def test_difficulty_range(self):
-        for tc in TEST_CASES:
-            assert 1 <= tc.difficulty <= 5, f"{tc.id} difficulty out of range"
-
-    def test_expected_tool_exists_in_schemas(self):
-        all_names = {t["function"]["name"] for t in TOOL_SCHEMAS}
-        for tc in TEST_CASES:
-            assert tc.expected_tool in all_names, (
-                f"{tc.id} expects '{tc.expected_tool}' which is not in TOOL_SCHEMAS"
-            )
-
-    def test_tools_subset_of_schemas(self):
-        all_names = {t["function"]["name"] for t in TOOL_SCHEMAS}
-        for tc in TEST_CASES:
-            for tool in tc.tools:
-                assert tool["function"]["name"] in all_names, (
-                    f"{tc.id} references unknown tool"
-                )
-
-    def test_unique_ids(self):
-        ids = [tc.id for tc in TEST_CASES]
-        assert len(ids) == len(set(ids)), f"Duplicate test IDs"
-
-
-class TestValidateToolCall(unittest.TestCase):
-    """Test the validation logic."""
-
-    def _make_response(self, tool_name, arguments):
-        return {
-            "choices": [{
-                "message": {
-                    "tool_calls": [{
-                        "type": "function",
-                        "function": {
-                            "name": tool_name,
-                            "arguments": json.dumps(arguments),
-                        },
-                    }]
-                }
-            }]
-        }
-
-    def test_exact_match_passes(self):
-        test = TEST_CASES[0]  # simple-read-1
-        resp = self._make_response("read_file", {"path": "/tmp/test.txt"})
-        result, tool, params, scores = validate_tool_call(resp, test)
-        assert result == TestResult.PASS
-        assert tool == "read_file"
-
-    def test_wrong_tool_fails(self):
-        test = TEST_CASES[0]
-        resp = self._make_response("terminal", {"command": "cat /tmp/test.txt"})
-        result, tool, params, scores = validate_tool_call(resp, test)
-        assert result == TestResult.FAIL
-
-    def test_no_tool_calls_fails(self):
-        test = TEST_CASES[0]
-        resp = {"choices": [{"message": {"content": "I'll read that file"}}]}
-        result, tool, params, scores = validate_tool_call(resp, test)
-        assert result == TestResult.FAIL
-
-    def test_partial_match_with_validators(self):
-        test = TEST_CASES[2]  # terminal-simple
-        resp = self._make_response("terminal", {"command": "ls -la"})
-        result, tool, params, scores = validate_tool_call(resp, test)
-        assert result == TestResult.PASS
-        assert scores.get("validator_command") is True
-
-    def test_validator_failure_is_partial(self):
-        test = TEST_CASES[2]  # terminal-simple, expects ls/dir/find
-        resp = self._make_response("terminal", {"command": "echo hello"})
-        result, tool, params, scores = validate_tool_call(resp, test)
-        # Tool matches but validator fails
-        assert result == TestResult.PARTIAL
-
-    def test_malformed_json_in_args(self):
-        test = TEST_CASES[0]
-        resp = {
-            "choices": [{
-                "message": {
-                    "tool_calls": [{
-                        "type": "function",
-                        "function": {
-                            "name": "read_file",
-                            "arguments": "{broken json",
-                        },
-                    }]
-                }
-            }]
-        }
-        result, tool, params, scores = validate_tool_call(resp, test)
-        assert result == TestResult.FAIL
-
-
-class TestDryRun(unittest.TestCase):
-    """Test the dry run mode."""
-
-    def test_dry_run_returns_all_tests(self):
-        results = run_dry_run()
-        assert len(results) == len(TEST_CASES)
-
-    def test_dry_run_all_skip(self):
-        results = run_dry_run()
-        for r in results:
-            assert r.result == TestResult.SKIP.value
-
-
-class TestReportGeneration(unittest.TestCase):
-    """Test report generation."""
-
-    def test_report_has_verdict(self):
-        results = [
-            TestRunResult(
-                test_id="test-1", category="simple", difficulty=1,
-                result="PASS", expected_tool="read_file", actual_tool="read_file",
-                expected_params={}, actual_params={},
-            ),
-        ]
-        report = generate_report(results, "test-model")
-        assert "VERDICT" in report
-        assert "VIABLE" in report
-        assert "test-model" in report
-
-    def test_report_pass_rate(self):
-        results = [
-            TestRunResult(test_id=f"t{i}", category="c", difficulty=1,
-                         result="PASS" if i < 3 else "FAIL",
-                         expected_tool="x", actual_tool="x",
-                         expected_params={}, actual_params={})
-            for i in range(5)
-        ]
-        report = generate_report(results, "m")
-        assert "60%" in report  # 3/5 = 60%
-
-
-@pytest.mark.skipif(
-    not os.environ.get("BONSAI_TOOL_CALL_URL"),
-    reason="No model server available (set BONSAI_TOOL_CALL_URL)",
-)
-class TestLiveInference:
-    """Live tests — requires a running model server."""
-
-    def test_server_responds(self):
-        import requests
-        url = os.environ["BONSAI_TOOL_CALL_URL"]
-        # Try a simple health check
-        resp = requests.get(url.replace("/chat/completions", "/models"), timeout=10)
-        assert resp.status_code in (200, 404)  # 404 is ok if endpoint differs
-
-    def test_simple_tool_call(self):
-        url = os.environ["BONSAI_TOOL_CALL_URL"]
-        model = os.environ.get("BONSAI_MODEL", "bonsai-1b")
-        result = _btc.run_test(TEST_CASES[0], url, model, timeout=60)
-        assert result.result in ("PASS", "PARTIAL")
-
-
-if __name__ == "__main__":
-    unittest.main()

tests/test_dflash_apple_silicon.py

@@ -0,0 +1,58 @@
+#!/usr/bin/env python3
+"""Tests for Apple Silicon DFlash benchmark planning helpers (issue #152)."""
+
+import os
+import sys
+from unittest.mock import patch
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(__file__)))
+
+from benchmarks.dflash_apple_silicon import (  # noqa: E402
+    build_mlx_benchmark_command,
+    detect_total_memory_gb,
+    render_report_template,
+    select_pair,
+)
+
+
+class TestPairSelection:
+    def test_prefers_qwen35_9b_on_36gb_mac(self):
+        pair = select_pair(total_memory_gb=36)
+        assert pair.slug == "qwen35-9b"
+        assert pair.base_model == "Qwen/Qwen3.5-9B"
+        assert pair.draft_model == "z-lab/Qwen3.5-9B-DFlash"
+
+    def test_falls_back_to_4b_when_memory_is_tight(self):
+        pair = select_pair(total_memory_gb=20)
+        assert pair.slug == "qwen35-4b"
+        assert pair.base_model == "Qwen/Qwen3.5-4B"
+
+
+class TestCommandGeneration:
+    def test_builds_upstream_mlx_benchmark_command(self):
+        pair = select_pair(total_memory_gb=36)
+        command = build_mlx_benchmark_command(pair, dataset="gsm8k", max_samples=64)
+        assert "python -m dflash.benchmark --backend mlx" in command
+        assert "--model Qwen/Qwen3.5-9B" in command
+        assert "--draft-model z-lab/Qwen3.5-9B-DFlash" in command
+        assert "--dataset gsm8k" in command
+        assert "--max-samples 64" in command
+        assert "--draft-sliding-window-size 4096" in command
+
+
+class TestReportTemplate:
+    def test_report_template_mentions_baseline_and_verdict(self):
+        pair = select_pair(total_memory_gb=36)
+        report = render_report_template(machine_label="M3 Max 36GB", pair=pair)
+        assert "DFlash Apple Silicon Benchmark Report" in report
+        assert "M3 Max 36GB" in report
+        assert "Qwen/Qwen3.5-9B" in report
+        assert "plain MLX or llama.cpp speculative decoding" in report
+        assert "Worth operationalizing locally?" in report
+
+
+class TestMemoryDetection:
+    @patch("benchmarks.dflash_apple_silicon.platform.system", return_value="Darwin")
+    @patch("benchmarks.dflash_apple_silicon.subprocess.check_output", return_value=b"38654705664\n")
+    def test_detect_total_memory_gb_on_macos(self, _mock_sysctl, _mock_system):
+        assert detect_total_memory_gb() == 36.0

tests/test_quant_selector.py

@@ -0,0 +1,177 @@
+#!/usr/bin/env python3
+"""Tests for quant_selector.py"""
+
+import sys
+import os
+import pytest
+from unittest.mock import patch, MagicMock
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(__file__)))
+from evolution.quant_selector import (
+    QuantLevel,
+    HardwareInfo,
+    QUANT_LEVELS,
+    detect_hardware,
+    estimate_kv_cache_gb,
+    estimate_model_memory_gb,
+    select_quant_level,
+)
+
+
+class TestQuantLevels:
+    def test_levels_keep_turboquant_quality_order_with_q4_fallback_last(self):
+        """TurboQuant levels should lead, with q4_0 reserved as the non-Turbo fallback."""
+        names = [level.name for level in QUANT_LEVELS]
+        assert names[:3] == ["turbo4", "turbo3", "turbo2"]
+        assert names[-1] == "q4_0"
+
+    def test_all_levels_have_required_fields(self):
+        for level in QUANT_LEVELS:
+            assert level.name
+            assert level.bits_per_channel > 0
+            assert level.compression_ratio > 1
+            assert level.quality_label
+            assert level.layer_adaptive >= 0
+            assert level.kv_type
+
+
+class TestKVEstimate:
+    def test_basic_estimate(self):
+        # 48 layers, 8 heads, 128 dim, 32K context, 3.5 bits
+        kv_gb = estimate_kv_cache_gb(32768, 48, 8, 128, 3.5)
+        assert kv_gb > 0
+        assert kv_gb < 10  # Should be reasonable
+
+    def test_longer_context_larger(self):
+        kv_32k = estimate_kv_cache_gb(32768, 48, 8, 128, 3.5)
+        kv_128k = estimate_kv_cache_gb(131072, 48, 8, 128, 3.5)
+        assert kv_128k > kv_32k
+
+    def test_higher_bits_larger(self):
+        kv_4b = estimate_kv_cache_gb(32768, 48, 8, 128, 4.0)
+        kv_2b = estimate_kv_cache_gb(32768, 48, 8, 128, 2.0)
+        assert kv_4b > kv_2b
+
+
+class TestHardwareDetection:
+    def test_detect_returns_info(self):
+        hw = detect_hardware()
+        assert hw.total_memory_gb > 0
+        assert hw.available_memory_gb > 0
+        assert hw.detection_method
+
+    @patch("evolution.quant_selector.platform.system", return_value="Linux")
+    @patch("builtins.open", create=True)
+    def test_linux_detection(self, mock_open, mock_system):
+        mock_open.return_value.__enter__().read.return_value = (
+            "MemTotal:       32000000 kB\n"
+            "MemAvailable:   24000000 kB\n"
+        )
+        hw = _detect_linux_fallback()
+        assert hw.total_memory_gb > 20
+
+
+def _detect_linux_fallback():
+    """Helper to test Linux detection with mocked /proc/meminfo."""
+    from evolution.quant_selector import _detect_linux
+    return _detect_linux()
+
+
+class TestSelection:
+    def test_selects_turbo4_for_large_memory(self):
+        """With plenty of memory, should pick turbo4 (best quality)."""
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=64,
+                available_memory_gb=48,
+                gpu_memory_gb=64,
+                gpu_name="Test GPU",
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=32768)
+            assert sel.level.name == "turbo4"
+            assert sel.headroom_gb > 0
+
+    def test_selects_smaller_for_tight_memory(self):
+        """With tight memory, should pick a smaller quant."""
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=16,
+                available_memory_gb=12,
+                gpu_memory_gb=16,
+                gpu_name="Test GPU",
+                cpu_cores=8,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=131072)
+            # Should pick a smaller quant for 128K context on 16GB
+            assert sel.level.bits_per_channel <= 4.0
+
+    def test_preferred_level(self):
+        """User can force a specific level."""
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=64,
+                available_memory_gb=48,
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(
+                model_size_gb=14.0, context_length=32768,
+                preferred_level="turbo2"
+            )
+            assert sel.level.name == "turbo2"
+
+    def test_env_vars_populated(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=64,
+                available_memory_gb=48,
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=32768)
+            assert "TURBO_LAYER_ADAPTIVE" in sel.env_vars
+            assert "-ctk" in sel.server_flags
+            assert "-ctv" in sel.server_flags
+
+    def test_warnings_on_low_headroom(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=18,
+                available_memory_gb=14,
+                gpu_memory_gb=18,
+                gpu_name="Test GPU",
+                cpu_cores=8,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=16.0, context_length=65536)
+            assert len(sel.warnings) > 0
+
+    def test_falls_back_to_turbo2_when_nothing_fits(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=8,
+                available_memory_gb=6,
+                gpu_memory_gb=8,
+                gpu_name="Tiny GPU",
+                cpu_cores=4,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=16.0, context_length=131072)
+            assert sel.level.name == "turbo2"
+
+    def test_reasoning_contains_key_info(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=32,
+                available_memory_gb=24,
+                is_apple_silicon=True,
+                chip_name="M4 Max",
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=32768)
+            assert "turbo4" in sel.reasoning
+            assert "M4 Max" in sel.reasoning or "32GB" in sel.reasoning

tests/test_tool_call_integration.py

@@ -0,0 +1,338 @@
+"""
+Integration test: turboquant compressed model passes hermes tool calls (issue #82).
+
+Validates that a TurboQuant-compressed model can:
+1. Parse hermes tool schemas correctly
+2. Format tool calls in OpenAI-compatible format
+3. Pass through the hermes agent conversation loop
+
+Tests are structured as contract tests -- they validate the schema/format
+compatibility without requiring a running model server. The live inference
+test is skipped by default (requires llama-server with TurboQuant model).
+
+Usage:
+    pytest tests/test_tool_call_integration.py -v
+    pytest tests/test_tool_call_integration.py -v -k live  # run live test if server available
+"""
+import json
+import os
+import pathlib
+import re
+import unittest
+
+import pytest
+
+ROOT = pathlib.Path(__file__).resolve().parents[1]
+PROFILE_PATH = ROOT / "profiles" / "hermes-profile-gemma4-turboquant.yaml"
+BENCHMARKS_DIR = ROOT / "benchmarks"
+
+
+class TestHermesProfileSchema(unittest.TestCase):
+    """Validate the hermes profile YAML has required fields for tool calling."""
+
+    @classmethod
+    def setUpClass(cls):
+        import yaml
+        cls.profile = yaml.safe_load(PROFILE_PATH.read_text())
+
+    def test_profile_has_providers(self):
+        assert "providers" in self.profile, "Profile must define providers"
+        assert "primary" in self.profile["providers"], "Must have primary provider"
+
+    def test_primary_provider_has_endpoint(self):
+        primary = self.profile["providers"]["primary"]
+        assert "endpoint" in primary, "Primary provider must have endpoint"
+        assert primary["endpoint"].startswith("http"), "Endpoint must be HTTP(S) URL"
+
+    def test_primary_provider_has_api_path(self):
+        primary = self.profile["providers"]["primary"]
+        assert "api_path" in primary, "Primary provider must have api_path"
+        assert "/chat/completions" in primary["api_path"], (
+            "api_path should be OpenAI-compatible /chat/completions"
+        )
+
+    def test_turboquant_settings_present(self):
+        primary = self.profile["providers"]["primary"]
+        assert "turboquant" in primary, "Must have turboquant config section"
+        tq = primary["turboquant"]
+        assert tq.get("enabled") is True, "TurboQuant must be enabled"
+        assert tq.get("kv_type") in ("turbo2", "turbo3", "turbo4"), (
+            "kv_type must be turbo2, turbo3, or turbo4"
+        )
+
+    def test_context_window_configured(self):
+        primary = self.profile["providers"]["primary"]
+        assert "context" in primary, "Must have context config"
+        ctx = primary["context"]
+        assert ctx.get("max_tokens", 0) >= 8192, (
+            "max_tokens should be >= 8192 for TurboQuant value proposition"
+        )
+
+
+class TestToolSchemaCompatibility(unittest.TestCase):
+    """Verify hermes tool schemas serialize to valid JSON for OpenAI tool_calls."""
+
+    SAMPLE_TOOL_SCHEMAS = [
+        {
+            "type": "function",
+            "function": {
+                "name": "read_file",
+                "description": "Read a text file with line numbers.",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "path": {"type": "string", "description": "File path"},
+                        "offset": {"type": "integer", "default": 1},
+                        "limit": {"type": "integer", "default": 500},
+                    },
+                    "required": ["path"],
+                },
+            },
+        },
+        {
+            "type": "function",
+            "function": {
+                "name": "execute_code",
+                "description": "Run a Python script.",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "code": {"type": "string", "description": "Python code"},
+                    },
+                    "required": ["code"],
+                },
+            },
+        },
+        {
+            "type": "function",
+            "function": {
+                "name": "web_search",
+                "description": "Search the web.",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "query": {"type": "string"},
+                        "max_results": {"type": "integer", "default": 5},
+                    },
+                    "required": ["query"],
+                },
+            },
+        },
+    ]
+
+    def test_tool_schemas_serialize_to_json(self):
+        """Tool schemas must serialize without errors."""
+        serialized = json.dumps(self.SAMPLE_TOOL_SCHEMAS)
+        assert len(serialized) > 0
+        parsed = json.loads(serialized)
+        assert len(parsed) == len(self.SAMPLE_TOOL_SCHEMAS)
+
+    def test_tool_schemas_have_required_openai_fields(self):
+        """Each tool schema must have the fields OpenAI expects."""
+        for tool in self.SAMPLE_TOOL_SCHEMAS:
+            assert tool["type"] == "function", "Tool type must be 'function'"
+            fn = tool["function"]
+            assert "name" in fn, "Function must have name"
+            assert "description" in fn, "Function must have description"
+            assert "parameters" in fn, "Function must have parameters"
+            params = fn["parameters"]
+            assert params["type"] == "object", "Parameters type must be 'object'"
+            assert "properties" in params, "Parameters must have properties"
+
+    def test_tool_call_response_format(self):
+        """Verify tool_call response matches OpenAI format."""
+        tool_call = {
+            "id": "call_abc123",
+            "type": "function",
+            "function": {
+                "name": "read_file",
+                "arguments": json.dumps({"path": "/tmp/test.txt"}),
+            },
+        }
+        args = json.loads(tool_call["function"]["arguments"])
+        assert args["path"] == "/tmp/test.txt"
+        assert tool_call["function"]["name"] in [
+            t["function"]["name"] for t in self.SAMPLE_TOOL_SCHEMAS
+        ]
+
+    def test_tool_names_are_valid_identifiers(self):
+        """Tool names must be valid Python identifiers for hermes dispatch."""
+        for tool in self.SAMPLE_TOOL_SCHEMAS:
+            name = tool["function"]["name"]
+            assert re.match(r"^[a-zA-Z_][a-zA-Z0-9_]*$", name), (
+                f"Tool name \'{name}\' is not a valid identifier"
+            )
+
+
+class TestTurboquantServerConfig(unittest.TestCase):
+    """Validate server startup configuration matches hermes profile."""
+
+    def test_server_command_has_turboquant_flags(self):
+        """The server command in the profile must include -ctk/-ctv flags."""
+        profile_text = PROFILE_PATH.read_text()
+        assert "-ctk" in profile_text, "Profile server command must include -ctk flag"
+        assert "-ctv" in profile_text, "Profile server command must include -ctv flag"
+
+    def test_server_command_has_context_flag(self):
+        """Server command must set context size."""
+        profile_text = PROFILE_PATH.read_text()
+        assert re.search(r"-c\s+\d+", profile_text), (
+            "Server command must include -c <context_size> flag"
+        )
+
+    def test_layer_adaptive_env_var(self):
+        """Profile must set TURBO_LAYER_ADAPTIVE env var."""
+        profile_text = PROFILE_PATH.read_text()
+        assert "TURBO_LAYER_ADAPTIVE" in profile_text, (
+            "Profile must configure TURBO_LAYER_ADAPTIVE"
+        )
+
+
+class TestBenchmarkData(unittest.TestCase):
+    """Validate benchmark test prompts include tool-call test cases."""
+
+    @classmethod
+    def setUpClass(cls):
+        prompts_path = BENCHMARKS_DIR / "test_prompts.json"
+        cls.prompts = json.loads(prompts_path.read_text())
+
+    def test_has_tool_call_test_prompt(self):
+        """Benchmark prompts must include a tool-call format test."""
+        categories = [p.get("category") for p in self.prompts]
+        assert "tool_call_format" in categories, (
+            "Benchmark must include a tool_call_format test case"
+        )
+
+    def test_tool_call_prompt_expects_json(self):
+        """Tool call test prompt must expect JSON in the response."""
+        tool_prompt = next(
+            p for p in self.prompts if p.get("category") == "tool_call_format"
+        )
+        pattern = tool_prompt.get("expected_pattern", "")
+        assert "json" in pattern.lower() or "\\{" in pattern, (
+            "Tool call prompt must expect JSON-formatted response"
+        )
+
+
+@pytest.mark.skipif(
+    not os.environ.get("TURBOQUANT_SERVER_URL"),
+    reason="No TurboQuant server available (set TURBOQUANT_SERVER_URL to run)",
+)
+class TestLiveToolCallIntegration:
+    """Live integration test -- requires running llama-server with TurboQuant."""
+
+    def test_server_health(self):
+        """Server must respond to /v1/models endpoint."""
+        import requests
+        url = os.environ["TURBOQUANT_SERVER_URL"]
+        resp = requests.get(f"{url}/v1/models", timeout=10)
+        assert resp.status_code == 200
+        data = resp.json()
+        assert "data" in data
+        assert len(data["data"]) > 0
+
+    def test_tool_call_completion(self):
+        """Model must return a valid tool_call for a read_file prompt."""
+        import requests
+        url = os.environ["TURBOQUANT_SERVER_URL"]
+        tools = [
+            {
+                "type": "function",
+                "function": {
+                    "name": "read_file",
+                    "description": "Read a file",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"path": {"type": "string"}},
+                        "required": ["path"],
+                    },
+                },
+            }
+        ]
+        resp = requests.post(
+            f"{url}/v1/chat/completions",
+            json={
+                "model": "gemma-4",
+                "messages": [
+                    {"role": "user", "content": "Read the file at /tmp/test.txt"}
+                ],
+                "tools": tools,
+                "tool_choice": "auto",
+            },
+            timeout=120,
+        )
+        assert resp.status_code == 200
+        data = resp.json()
+        choice = data["choices"][0]
+        msg = choice["message"]
+        if "tool_calls" in msg and msg["tool_calls"]:
+            tc = msg["tool_calls"][0]
+            assert tc["type"] == "function"
+            assert tc["function"]["name"] == "read_file"
+            args = json.loads(tc["function"]["arguments"])
+            assert "path" in args
+        else:
+            assert len(msg.get("content", "")) > 0
+
+    def test_tool_call_with_multiple_tools(self):
+        """Model must handle multiple available tools."""
+        import requests
+        url = os.environ["TURBOQUANT_SERVER_URL"]
+        tools = [
+            {
+                "type": "function",
+                "function": {
+                    "name": "read_file",
+                    "description": "Read a file",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"path": {"type": "string"}},
+                        "required": ["path"],
+                    },
+                },
+            },
+            {
+                "type": "function",
+                "function": {
+                    "name": "web_search",
+                    "description": "Search the web",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"query": {"type": "string"}},
+                        "required": ["query"],
+                    },
+                },
+            },
+            {
+                "type": "function",
+                "function": {
+                    "name": "execute_code",
+                    "description": "Run Python code",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"code": {"type": "string"}},
+                        "required": ["code"],
+                    },
+                },
+            },
+        ]
+        resp = requests.post(
+            f"{url}/v1/chat/completions",
+            json={
+                "model": "gemma-4",
+                "messages": [
+                    {"role": "user", "content": "Search the web for 'bitcoin price'"}
+                ],
+                "tools": tools,
+                "tool_choice": "auto",
+            },
+            timeout=120,
+        )
+        assert resp.status_code == 200
+        data = resp.json()
+        assert "choices" in data
+        assert len(data["choices"]) > 0
+
+
+if __name__ == "__main__":
+    unittest.main()