GPT OSS 20B High
by OpenAI
GPT OSS 20B High is a model from OpenAI, released on August 5, 2025. It has a 131K-token context window and can generate up to 131K output tokens. API pricing is $0.1 per 1M input tokens and $0.5 per 1M output tokens. It ranks #91 of 220 models by Intelligence Score with a score of 27.6.
GPT-OSS-20B High provides enhanced reasoning capabilities with high-effort thinking for complex problems. This variant offers deeper analysis and more thorough responses compared to the base model, making it ideal for challenging tasks that require extended reasoning. Achieves near-parity with OpenAI o4-mini on core reasoning benchmarks while running efficiently on a single 80 GB GPU, or on edge devices with just 16 GB of memory. Note: While referred to as '20b' for simplicity, it technically has 20.9B parameters.
Key specs
| Provider | OpenAI |
|---|---|
| Release date | 2025-08-05 |
| License | open source |
| Context window | 131K tokens |
| Max output | 131K tokens |
| Input price | $0.1 / 1M tokens |
| Output price | $0.5 / 1M tokens |
| Intelligence Score | 27.6 |
Quick answers
What is the context window of GPT OSS 20B High?
GPT OSS 20B High has a context window of 131K tokens, with a maximum output of 131K tokens per response.
How much does GPT OSS 20B High cost?
GPT OSS 20B High costs $0.1 per 1M input tokens and $0.5 per 1M output tokens through the API.
Is GPT OSS 20B High open source?
Yes. GPT OSS 20B High is an open model released under the open source license.
When was GPT OSS 20B High released?
OpenAI released GPT OSS 20B High on August 5, 2025.
How does GPT OSS 20B High compare to Claude Opus 5.5 (Reasoning)?
GPT OSS 20B High scores 27.6 on the Intelligence Score versus 60.5 for Claude Opus 5.5 (Reasoning), the current #1. Input pricing is $0.1 vs $4 per 1M tokens.
Compare on the AI leaderboard · All AI models
More models from OpenAI
- GPT-6 Astra (Reasoning) — OpenAI
- GPT-6.1 Sol (Reasoning) — OpenAI
- GPT-5.6 Terra (Reasoning) — OpenAI
- GPT-6 Sol (Reasoning) — OpenAI
- GPT-5.5 Pro (Reasoning) — OpenAI
- GPT-6 Luna (Reasoning) — OpenAI