When was gpt-oss-20b released?

gpt-oss-20b was released by OpenAI on Tuesday, Aug 5 2025.

Who made gpt-oss-20b?

gpt-oss-20b was built by OpenAI. Creators of ChatGPT and the GPT series of models. Pioneered large-scale language model research.

What benchmark scores did gpt-oss-20b get?

gpt-oss-20b reports 5 tracked benchmark scores — SWE-Bench Verified: 60.7%; Humanity's Last Exam (no tools): 10.9%; Humanity's Last Exam (with tools): 17.3%; GPQA Diamond: 71.5%; MMLU: 85.3%. Scores are the figures published at release by OpenAI.

What is the context window of gpt-oss-20b?

gpt-oss-20b has a context window of 128k. That is the maximum amount of input plus output the model can hold in a single request.

How many parameters does gpt-oss-20b have?

gpt-oss-20b is reported at 21B parameters.

Is gpt-oss-20b open source?

Partly. gpt-oss-20b is an open-weight model: the trained weights are free to download and run locally or on your own infrastructure, but the training data and code are not fully released and the license may restrict some commercial uses. It is not open source in the strict sense.

What came before and after gpt-oss-20b?

OpenAI's previous tracked release was gpt-oss-120b on Aug 5 2025. It was followed by GPT-5 on Aug 7 2025.

gpt-oss-20bOpen Weight

Released Aug 5 2025

gpt-oss-20b is an AI model released by OpenAI on Tuesday, Aug 5 2025, the same day as gpt-oss-120b. It is an open-weight model — the trained weights are available to download and run. It is a 21B parameter model with a 128k token context window. Benchmark results (shown below) cover SWE-Bench Verified, Humanity's Last Exam, GPQA Diamond, and MMLU.

Available from


AkashML	—	$0.02	$0.10	128K	fp4
Darkbloom	—	$0.02	$0.10	128K	fp8
CoreWeave	—	$0.03	$0.13	128K	fp4
DekaLLM	—	$0.029	$0.14	128K	bf16
DeepInfra	—	$0.03	$0.14	128K	bf16
Parasail	—	$0.03	$0.15	128K	fp4
Novita	—	$0.04	$0.15	128K	fp4
Phala	—	$0.04	$0.15	128K	—
Amazon Bedrock	—	$0.07	$0.15	128K	—
Amazon Bedrock	eu-west-1	$0.07	$0.15	128K	—
SiliconFlow	—	$0.04	$0.18	128K	fp8
Together	—	$0.05	$0.20	128K	—
Google	us-central1	$0.07	$0.25	128K	—
Groq	—	$0.075	$0.30	128K	—

USD per 1M tokens

Benchmarks

Coding

SWE-Bench Verified

60.7%

#46 of 57

Reasoning & science

Humanity's Last Exam

10.9%

no tools

17.3%

with tools

no tools

#22 of 22

with tools

#42 of 42

GPQA Diamond

71.5%

#48 of 59

MMLU

85.3%

#3 of 4

Compare gpt-oss-20b with

gpt-oss-20b

Suggested comparisons

gpt-oss-20bvsGPT-6 Astra gpt-oss-20bvsClaude Fable 5.1 gpt-oss-20bvsGemini 3.8 Flash gpt-oss-20bvsMuse Spark 1.3 gpt-oss-20bvsGrok 4.6 gpt-oss-20bvsDeepSeek-V4.1-Flash gpt-oss-20bvsMistral Medium 3.5 gpt-oss-20bvsKimi K3 gpt-oss-20bvsGLM-5.3-Flash gpt-oss-20bvsQwen3.8-Max-0902 gpt-oss-20bvsNemotron 3.5 Lightning

Frequently asked questions

: gpt-oss-20b was released by OpenAI on Tuesday, Aug 5 2025.
: gpt-oss-20b was built by OpenAI. Creators of ChatGPT and the GPT series of models. Pioneered large-scale language model research.
: gpt-oss-20b reports 5 tracked benchmark scores — SWE-Bench Verified: 60.7%; Humanity's Last Exam (no tools): 10.9%; Humanity's Last Exam (with tools): 17.3%; GPQA Diamond: 71.5%; MMLU: 85.3%. Scores are the figures published at release by OpenAI.
: gpt-oss-20b has a context window of 128k. That is the maximum amount of input plus output the model can hold in a single request.
: gpt-oss-20b is reported at 21B parameters.
: Partly. gpt-oss-20b is an open-weight model: the trained weights are free to download and run locally or on your own infrastructure, but the training data and code are not fully released and the license may restrict some commercial uses. It is not open source in the strict sense.
: OpenAI's previous tracked release was gpt-oss-120b on Aug 5 2025. It was followed by GPT-5 on Aug 7 2025.