GPT OSS 20B
Latest GPT OSSOpen GPT reasoning model for self-hosted agents and controllable deployments
- Strongest at
- Math Better than 46% of the past year's models · one benchmark so far
- Price per 1M tokens
- $0.018 / $0.09 input / output · OpenRouter
- Context window
- 131.1K tokens in · 32.8K out
Capability profile
Scored against 91 models released in the past year. 50 is the average model; further out is better.
No results yet: Coding, Reading & documents, Reasoning, Visual reasoning, Computer use
Where these scores come from
Each capability combines the published benchmarks listed here, rescaled so 50 is the reference average and 15 points is one standard deviation. A hollow dot rests on a single benchmark.
- Writing EQ-Bench Creative Writing v3, EQ-Bench Longform Writing
- Math OTIS Mock AIME 2024-2025
- Science GPQA diamond, WeirdML v2
Price against capability
The shaded corner holds anything better and cheaper. Prices blend input and output 3 to 1.
Nothing from the past year scores higher on Math for less.
Nothing from the past year scores higher on Writing for less.
Nothing from the past year scores higher on Science for less.
OpenRouter $0.018 / $0.09 · Cached input $0.009
The GPT OSS line
2 releases in the GPT OSS line.
Epoch benchmark boards
Best result on each board it has run. The tick marks the leader.
OpenAI's past year
24 release days in the past year, about one every 15 days. The next is expected around Nov 8, 2026. Forecasts ↗
Specifications
Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.