Model benchmarks
openai/gpt-oss-120b local LLM performance
As of August 2026, openai/gpt-oss-120b runs at up to 176.4 tok/s for local inference (best of 2 community benchmark runs across 2 GPUs).
Model size
120B
Peak speed
176.4 tok/s
Average speed
111.8 tok/s
Min memory
59.0 GB
Max context
65,536 tokens
Avg output / run
9,119 tokens
Avg runtime / run
3m 31s
Avg quality
79.9
Benchmark runs
2
GPUs tested
2
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 | 79.7 | 79.9 | 80.2 |
| Agent Workflow | 83.7 | 86.3 | 88.8 |
| Code Generation | 74.0 | 74.1 | 74.2 |
| Role Play & Narrative | 78.6 | 80.7 | 82.7 |
| Research & Analysis | 78.4 | 78.8 | 79.1 |
Performance by hardware and tool
Every hardware/tool/quantization combination openai/gpt-oss-120b 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 |
|---|---|---|---|---|---|---|---|---|
| NVIDIA L4 | Ollama | MXFP4 | 176.4 tok/s | 176.4 tok/s | 60.0 GB | 65,536 tokens | 79.7 | 1 |
| Apple M2 Max | LM Studio | — | 47.1 tok/s | 47.1 tok/s | 59.0 GB | 8,192 tokens | 80.2 | 1 |
Frequently asked questions
- Is gpt-oss:120b good for coding?
- In our benchmarks, gpt-oss:120b scores 74.1/100 for coding. It runs at about 111.8 tok/s, so if you want more speed, Gemma4:E2B/QAT-MTP@131K is faster (~303.9 tok/s) and still scores well for coding (66.9/100).
- Is gpt-oss:120b good for agentic (tool-using) tasks?
- In our benchmarks, gpt-oss:120b scores 86.3/100 for agentic workflows. It runs at about 111.8 tok/s, so if you want more speed, mudler/Ornith-1.5-35B-A3B-APEX-MTP-GGUF is faster (~190.9 tok/s) and still scores well for agentic workflows (85.7/100).
- How fast is openai/gpt-oss-120b for local inference?
- Across 2 community benchmark runs, openai/gpt-oss-120b reaches up to 176.4 tok/s and averages 111.8 tok/s, with the fastest results on NVIDIA L4.
- How much memory does openai/gpt-oss-120b need?
- The leanest observed configuration used about 59.0 GB of memory (quantizations tested: MXFP4).
- Which tools have been used to run openai/gpt-oss-120b?
- Benchmarks were submitted using LM Studio, Ollama. Results are community-contributed and updated as new runs arrive.