Model benchmarks
Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit local LLM performance
As of August 2026, Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit runs at up to 12.0 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
40B
Peak speed
12.0 tok/s
Average speed
10.1 tok/s
Min memory
40.3 GB
Max context
65,536 tokens
Avg runtime / run
24m 43s
Avg quality
75.1
Benchmark runs
2
GPUs tested
1
Quality by task
Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 2 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.
| Task | P5 (low) | Avg | P95 (high) |
|---|---|---|---|
| Overall | 73.9 | 75.1 | 76.3 |
| Agent Workflow | 71.9 | 73.2 | 74.4 |
| Code Generation | 69.6 | 71.3 | 72.9 |
| Role Play & Narrative | 78.1 | 80.0 | 81.9 |
| Research & Analysis | 72.2 | 76.1 | 80.0 |
Performance by hardware and tool
Every hardware/tool/quantization combination Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit has been benchmarked on, ranked by peak token generation speed. Last updated August 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| Apple M5 Max | oMLX | — | 12.0 tok/s | 10.1 tok/s | 40.3 GB | 65,536 tokens | 75.1 | 2 |
Benchmark runs
All 2 Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit good for coding?
- In our benchmarks, Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit scores 71.3/100 for coding. It runs at about 10.1 tok/s, so if you want more speed, IsValorum/Nex-N2.5-mini-APEX-I-MiniPlus-V2.1-GGUF:1 is faster (~25.7 tok/s) and still scores well for coding (85.0/100).
- Is Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit good for agentic (tool-using) tasks?
- In our benchmarks, Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit scores 73.2/100 for agentic workflows. It runs at about 10.1 tok/s, so if you want more speed, granite-4.2-8b-Q4_K_L is faster (~71.6 tok/s) and still scores well for agentic workflows (92.9/100).
- How fast is Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit for local inference?
- Across 2 community benchmark runs, Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit reaches up to 12.0 tok/s and averages 10.1 tok/s, with the fastest results on Apple M5 Max.
- How much memory does Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit need?
- The leanest observed configuration used about 40.3 GB of memory.
- Which tools have been used to run Qwen3.6-40B-Claude-4.6-Opus-Deckard-Heretic-Uncensored-Thinking-8bit?
- Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.