gpt oss 20b
openai/gpt-oss-20b
Open Source · chat · open-weights
Context
131.1K
Max output
32.8K
Input $/1M
$0.03
Output $/1M
$0.13
Modalities
text
Released
04 Aug 2025
License: apache-2.0 · openai/gpt-oss-20b
AI summary
● machine-written
OpenAI releases gpt-oss-20b open-weight model under Apache 2.0
GPT OSS 20b is a 20-billion parameter open-weight model released by OpenAI using a Mixture-of-Experts architecture with 3.6B active parameters per forward pass. The model supports a 131K token context window and is available through multiple providers at varying price points. It is trained for chat and supports reasoning, function calling, tool use, and structured outputs.
What's new
- Released August 5, 2025 under Apache 2.0 license
- 131,072 token context window with 32,768 token maximum output
- Mixture-of-Experts architecture optimized for consumer/single-GPU deployment
- Available from 12 providers with pricing starting at $0.029 per million input tokens
Best for
Deployment on consumer or single-GPU hardwareLow-latency inference applicationsAgentic systems requiring function calling and tool useReasoning tasks with configurable reasoning levels