---
id: "openai/gpt-oss-20b"
also: ["openai/gpt-oss-20b", "hf/openai/gpt-oss-20b", "requesty/openai/gpt-oss-20b", "vercel/openai/gpt-oss-20b"]
cheapest_list_usd_per_m: "0.018 in, 0.09 out, at akashml/fp4 on openrouter"
cheapest_effective_usd_per_m: "0.0151 in, 0.0888 out, at darkbloom/fp8 on openrouter"
context: 131072
max_output: 117964
input: ["text"]
output: ["text"]
released: "2025-08-05"
reasoning: ["high", "medium", "low"]
api: ["/api/v1/chat/completions", "/api/v1/completions", "/api/v1/messages", "/api/v1/responses"]
params: ["frequency_penalty", "include_reasoning", "logit_bias", "logprobs", "max_tokens", "min_p", "presence_penalty", "reasoning", "reasoning_effort", "repetition_penalty", "response_format", "seed", "stop", "structured_outputs", "temperature", "tool_choice", "tools", "top_k", "top_logprobs", "top_p"]
endpoints: 34
gateways: "openrouter (16), hf (8), vercel (8), requesty (2)"
no_training: 25
trains: 0
retains_prompts: 2
coding: 20.7000
agentic: 1.2000
---
# OpenAI: gpt-oss-20b

OpenAI's gpt-oss-20b is a 21-billion parameter, open-weight reasoning model released under the Apache 2.0 license. Designed for local execution on consumer hardware and low-latency workflows, it delivers reasoning capabilities and benchmark performance comparable to the OpenAI o3-mini models.