gpt-oss-20b
gpt-oss-20b is the smaller downloadable gpt-oss release. Its 21 billion total parameters include about 3.6 billion active parameters, and its license is Apache 2.0.
Local customization is a deployment choice
The model accepts and returns text, supports configurable reasoning and can be fine-tuned. Its documented window is 131,072 tokens. OpenAI positions it for local and specialized workloads; the weights do not include a managed application service.gpt-oss-20b official model documentation
Evaluate the complete local application
Practical guidance. For a private document assistant, measure retrieval, prompt length and generated answers together. Confirm the serving software handles the model response format. Compare resource use under realistic concurrent requests, rather than using parameter count as a latency estimate.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | gpt-oss-20b |
| Context window | 131,072 tokens Published model context; input and output share this budget gpt-oss-20b official model documentation |
| Input and output | text in; text outgpt-oss-20b official model documentation |
| License | Apache 2.0gpt-oss-20b official model documentation |
Related models
- gpt-oss-120b
OpenAI ยท gpt-oss
gpt-oss-120b is a downloadable text reasoning model with 117 billion total parameters and about 5.1 billion active parameters.
Launch date not confirmedDownloadable