Model benchmarks
gpt-oss-20b-UD-Q6_K_XL local LLM performance
As of September 2026, gpt-oss-20b-UD-Q6_K_XL runs at up to 179.0 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
20B
Peak speed
179.0 tok/s
Average speed
151.2 tok/s
Avg prefill
1658.7 tok/s
Min memory
9.8 GB
Max context
131,072 tokens
Avg output / run
10,450 tokens
Avg runtime / run
1m 17s
Avg quality
74.0
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 | 70.8 | 74.0 | 77.2 |
| Agent Workflow | 76.0 | 76.2 | 76.4 |
| Code Generation | 69.5 | 74.1 | 78.7 |
| Role Play & Narrative | 63.4 | 68.6 | 73.7 |
| Research & Analysis | 73.9 | 77.1 | 80.2 |
Performance by hardware and tool
Every hardware/tool/quantization combination gpt-oss-20b-UD-Q6_K_XL has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 5060 | llama.cpp | Q6_K | 179.0 tok/s | 151.2 tok/s | 9.8 GB | 131,072 tokens | 74.0 | 2 |
Frequently asked questions
- How fast is gpt-oss-20b-UD-Q6_K_XL for local inference?
- Across 2 community benchmark runs, gpt-oss-20b-UD-Q6_K_XL reaches up to 179.0 tok/s and averages 151.2 tok/s, with the fastest results on NVIDIA GeForce RTX 5060.
- How much memory does gpt-oss-20b-UD-Q6_K_XL need?
- The leanest observed configuration used about 9.8 GB of memory (quantizations tested: Q6_K).
- Which tools have been used to run gpt-oss-20b-UD-Q6_K_XL?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.