gpt-oss-120b
gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases.
Specifications
| Lab | OpenAI |
|---|---|
| Context window | 131.1K tokens |
| Input price | $0.15/1M |
| Output price | $0.60/1M |
| Release | Aug 2025 |
Strengths
- First open-weight release from OpenAI (Apache 2.0)
- Efficient 117B MoE architecture (5.1B active params)
- Strong reasoning and agentic capabilities
- Ultra-budget pricing for self-hosted deployment
Weaknesses
- Small active parameter count limits raw power
- No multimodal/vision input support
- Less polished than closed OpenAI models
Best for
- Self-hosted production deployments
- Cost-efficient agentic workflows
- Research and fine-tuning
- Budget-conscious enterprise use
In Depth: gpt-oss-120b
Summary
gpt-oss-120b is an AI model from OpenAI.
Released in Aug 2025. It currently appears in the Overall category on LMRank and 1 other category. It takes text input and produces text output, with a context window of 131.1K tokens. Input pricing is $0.15 per 1M tokens and output is $0.60 per 1M tokens on OpenRouter.