gpt oss 120b
openai/gpt-oss-120b
Open Source · chat · open-weights
Context
131.1K
Max output
131.1K
Input $/1M
$0.04
Output $/1M
$0.10
Modalities
text
Released
04 Aug 2025
License: apache-2.0 · openai/gpt-oss-120b
AI summary
● machine-written
OpenAI releases gpt-oss-120b open-weights model
gpt-oss-120b is a 117B-parameter Mixture-of-Experts language model released by OpenAI as open-weights, designed for reasoning, agentic, and general-purpose production use cases. The model activates 5.1B parameters per forward pass and supports a 131K token context window. It is available through multiple providers at varying price points and includes native tool use and chain-of-thought capabilities.
What's new
- Open-weights 117B MoE model with 5.1B active parameters per forward pass
- 131K token context window with support for configurable reasoning depth
- Native MXFP4 quantization optimized for single H100 GPU deployment
- Supports function calling, browsing, and structured output generation
Best for
High-reasoning tasks and agentic workflowsProduction use cases requiring tool use and function callingGeneral-purpose language tasks with extended context requirements