Is gpt-oss-120b-Sonnet-Reasoning-Distilled good for coding?

In our benchmarks, gpt-oss-120b-Sonnet-Reasoning-Distilled scores 67.9/100 for coding. It runs at about 82.3 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).

Is gpt-oss-120b-Sonnet-Reasoning-Distilled good for agentic (tool-using) tasks?

In our benchmarks, gpt-oss-120b-Sonnet-Reasoning-Distilled scores 74.7/100 for agentic workflows. It runs at about 82.3 tok/s, so if you want more speed, Ornith-1.5-35B-A3B-BigBang-oQ8e-mtp is faster (~102.7 tok/s) and still scores well for agentic workflows (91.1/100).

How fast is gpt-oss-120b-Sonnet-Reasoning-Distilled for local inference?

Across 10 community benchmark runs, gpt-oss-120b-Sonnet-Reasoning-Distilled reaches up to 82.9 tok/s and averages 82.3 tok/s, with the fastest results on Apple M4 Max.

How much memory does gpt-oss-120b-Sonnet-Reasoning-Distilled need?

The leanest observed configuration used about 59.9 GB of memory.

Which tools have been used to run gpt-oss-120b-Sonnet-Reasoning-Distilled?

Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.

Model benchmarks

gpt-oss-120b-Sonnet-Reasoning-Distilled local LLM performance

As of October 2026, gpt-oss-120b-Sonnet-Reasoning-Distilled runs at up to 82.9 tok/s for local inference (best of 10 community benchmark runs across 1 GPU).

oMLX

ShareReddit X

Model size

120B

Peak speed

82.9 tok/s

Average speed

82.3 tok/s

Avg PP

1040.7 tok/s

Min memory

59.9 GB

Max context

65,536 tokens

Avg output / run

7,600 tokens

Avg runtime / run

1m 39s

Avg quality

70.8

Benchmark runs

GPUs tested

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 10 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	66.0	70.8	75.1
Agent Workflow	65.4	74.7	80.5
Code Generation	58.8	67.9	74.8
Role Play & Narrative	55.3	70.6	78.0
Research & Analysis	60.3	69.9	79.2

Performance by hardware and tool

Every hardware/tool/quantization combination gpt-oss-120b-Sonnet-Reasoning-Distilled has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.

Hardware	Tool	Quant	Best tok/s	Avg tok/s	Memory	Context	Quality	Runs
Apple M4 Max	oMLX	—	82.9 tok/s	82.3 tok/s	59.9 GB	65,536 tokens	70.8	10

Benchmark runs

All 10 gpt-oss-120b-Sonnet-Reasoning-Distilled runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

Is gpt-oss-120b-Sonnet-Reasoning-Distilled good for coding?: In our benchmarks, gpt-oss-120b-Sonnet-Reasoning-Distilled scores 67.9/100 for coding. It runs at about 82.3 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).
Is gpt-oss-120b-Sonnet-Reasoning-Distilled good for agentic (tool-using) tasks?: In our benchmarks, gpt-oss-120b-Sonnet-Reasoning-Distilled scores 74.7/100 for agentic workflows. It runs at about 82.3 tok/s, so if you want more speed, Ornith-1.5-35B-A3B-BigBang-oQ8e-mtp is faster (~102.7 tok/s) and still scores well for agentic workflows (91.1/100).
How fast is gpt-oss-120b-Sonnet-Reasoning-Distilled for local inference?: Across 10 community benchmark runs, gpt-oss-120b-Sonnet-Reasoning-Distilled reaches up to 82.9 tok/s and averages 82.3 tok/s, with the fastest results on Apple M4 Max.
How much memory does gpt-oss-120b-Sonnet-Reasoning-Distilled need?: The leanest observed configuration used about 59.9 GB of memory.
Which tools have been used to run gpt-oss-120b-Sonnet-Reasoning-Distilled?: Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.