Model benchmarks
coder local LLM performance
As of November 2026, coder runs at up to 57.5 tok/s for local inference (best of 4 community benchmark runs across 1 GPU).
Model size
Unknown
Peak speed
57.5 tok/s
Average speed
42.3 tok/s
Avg PP
553.6 tok/s
Min memory
19.6 GB
Max context
65,536 tokens
Avg output / run
24,972 tokens
Avg runtime / run
13m 10s
Avg quality
72.5
Benchmark runs
4
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 4 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 | 67.1 | 72.5 | 75.6 |
| Agent Workflow | 36.5 | 68.2 | 91.2 |
| Code Generation | 28.7 | 55.0 | 71.4 |
| Role Play & Narrative | 83.9 | 85.8 | 86.8 |
| Research & Analysis | 73.1 | 81.0 | 87.0 |
Performance by hardware and tool
Every hardware/tool/quantization combination coder has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M | OpenAI-compatible | — | 57.5 tok/s | 42.3 tok/s | 19.6 GB | 65,536 tokens | 72.5 | 4 |
Benchmark runs
All 4 coder runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is coder good for coding?
- In our benchmarks, coder scores 55.0/100 for coding. It runs at about 42.3 tok/s, so if you want more speed, nvidia/Qwen3.8-27B-NVFP4 is faster (~63.3 tok/s) and still scores well for coding (82.7/100).
- Is coder good for agentic (tool-using) tasks?
- In our benchmarks, coder scores 68.2/100 for agentic workflows. It runs at about 42.3 tok/s, so if you want more speed, swift-iq2_xs is faster (~91.5 tok/s) and still scores well for agentic workflows (91.4/100).
- How fast is coder for local inference?
- Across 4 community benchmark runs, coder reaches up to 57.5 tok/s and averages 42.3 tok/s, with the fastest results on AMD Radeon RX 7900 XT/7900 XTX/7900 GRE/7900M.
- How much memory does coder need?
- The leanest observed configuration used about 19.6 GB of memory.
- Which tools have been used to run coder?
- Benchmarks were submitted using OpenAI-compatible. Results are community-contributed and updated as new runs arrive.