Model benchmarks

gpt-oss-20b-Q4_K_M local LLM performance

As of September 2026, gpt-oss-20b-Q4_K_M runs at up to 142.0 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

llama.cppQ4_K_M
ShareRedditX

Model size

20B

Peak speed

142.0 tok/s

Average speed

132.8 tok/s

Avg prefill

2090.4 tok/s

Min memory

9.8 GB

Max context

131,072 tokens

Avg output / run

5,348 tokens

Avg runtime / run

42s

Avg quality

64.9

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.

TaskP5 (low)AvgP95 (high)
Overall62.064.967.7
Agent Workflow74.679.484.1
Code Generation53.057.862.5
Role Play & Narrative58.359.260.0
Research & Analysis61.863.164.5

Performance by hardware and tool

Every hardware/tool/quantization combination gpt-oss-20b-Q4_K_M has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
NVIDIA GeForce RTX 5060llama.cppQ4_K_M142.0 tok/s132.8 tok/s9.8 GB131,072 tokens64.92

Frequently asked questions

Is gpt-oss-20b-Q4_K_M good for coding?
In our benchmarks, gpt-oss-20b-Q4_K_M scores 57.8/100 for coding. It runs at about 132.8 tok/s, so if you want more speed, Qwen3.6-35B-A3B-Q4_K_XL is faster (~156.9 tok/s) and still scores well for coding (82.2/100).
Is gpt-oss-20b-Q4_K_M good for agentic (tool-using) tasks?
In our benchmarks, gpt-oss-20b-Q4_K_M scores 79.4/100 for agentic workflows. It runs at about 132.8 tok/s, so if you want more speed, Tiel-Coder-35B-A3B-Q4_K_S is faster (~165.0 tok/s) and still scores well for agentic workflows (89.3/100).
How fast is gpt-oss-20b-Q4_K_M for local inference?
Across 2 community benchmark runs, gpt-oss-20b-Q4_K_M reaches up to 142.0 tok/s and averages 132.8 tok/s, with the fastest results on NVIDIA GeForce RTX 5060.
How much memory does gpt-oss-20b-Q4_K_M need?
The leanest observed configuration used about 9.8 GB of memory (quantizations tested: Q4_K_M).
Which tools have been used to run gpt-oss-20b-Q4_K_M?
Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.