gpt-oss-20b
OpenAI · Released Aug 2025
An open-weight 20-billion-parameter model from OpenAI using Mixture-of-Experts architecture with approximately 3.6 billion active parameters per forward pass.
- Strengths
- Offers efficient inference through sparse activation while maintaining a large context window of 131k tokens, balancing capability with computational efficiency.
- Best for
- Workloads where model weight availability and self-hosting matter, or batch processing where inference speed and resource consumption must stay low.
- Limitations
- Smaller and less capable than the full GPT-5 family models released in the same timeframe; gpt-oss-120b offers substantially more capacity if latency constraints permit.
Input / 1M
$0.018
Output / 1M
$0.09
Cached input / 1M
$0.009
Context window
131K
Price history
Input (solid)Output (dashed)
Price change
- 30d
- in decreased 0.0% out decreased 0.0%
- 90d
- in decreased 40.0% out decreased 30.8%
- 1y
- in decreased 37.9% out decreased 35.7%
- Since launch
- in decreased 37.9% out decreased 35.7%
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 29 Sep 2026 | $0.018 | $0.09 | $0.009 | Imported from OpenRouter | openrouter.ai |
| 22 Sep 2026 | $0.018 | $0.09 | — | Imported from OpenRouter | openrouter.ai |
| 1 Aug 2026 | $0.03 | $0.13 | $0.03 | Imported from OpenRouter | openrouter.ai |
| 31 Jul 2026 | $0.03 | $0.14 | — | Imported from OpenRouter | openrouter.ai |
| 29 Jul 2026 | $0.03 | $0.13 | $0.03 | Imported from OpenRouter | openrouter.ai |
| 26 Jul 2026 | $0.03 | $0.14 | — | Imported from OpenRouter | openrouter.ai |
| 25 Jul 2026 | $0.03 | $0.13 | $0.03 | Imported from OpenRouter | openrouter.ai |
| 24 Jul 2026 | $0.029 | $0.14 | — | Imported from OpenRouter | openrouter.ai |
| 15 Jul 2026 | $0.03 | $0.13 | $0.03 | Imported from OpenRouter | openrouter.ai |
| 11 Jun 2026 | $0.029 | $0.14 | — | Imported from OpenRouter | openrouter.ai |