GPT

OpenAI: gpt-oss-20b

openai/gpt-oss-20b

Descriptionfrom OpenRouter

gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...

Context length

131,072 tok

Max completion

117,964 tok

Knowledge cutoff

2024-06-30

Expires

—

Added to OpenRouter

8/5/2025

Cache last refreshed

11/12/2026

Primary modality

text->text

Source

OpenRouter

Input modalities

text

Output modalities

text

Pricing

Field	Per token (USD)	Per million
prompt	0.000000018	$0.018
completion	0.00000009	$0.090
input_cache_read	0.000000009	$0.009

Supported parameters(20)

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p

Default parameters

{
  "temperature": null,
  "top_p": null,
  "frequency_penalty": null
}

OpenRouter page Hugging Face

Videos

47:51

Ranking major U.S. Open-Weights Models. (Where does Ripple 65B fit in?)

Nov 6, 2026 · YouTube

Rounds

10 answersin this database. A round opens every model's answer to the same prompt.

Nov 5, 2026Small ModelsAWS Quick Ten57mediocrein a video round answer
Nov 5, 2026Small ModelsRust Quick Ten53mediocrein a video round answer
Nov 5, 2026Small ModelsAWS Quick Ten40mediocrein a video round answer
Nov 5, 2026Small ModelsRust Quick Ten67goodin a video round answer
Nov 4, 2026Small ModelsTailwind Quick Ten93excellentin a video round answer
Nov 4, 2026Small ModelsPython Quick Ten97excellentin a video round answer
Nov 4, 2026Small ModelsPython Quick Ten90excellentin a video round answer
Nov 4, 2026Small ModelsTailwind Quick Ten90excellentin a video round answer
Nov 4, 2026Small ModelsTailwind Quick Ten97excellentin a video round answer
Nov 4, 2026Small ModelsPython Quick Ten97excellentin a video round answer