New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
gpt-oss-20b
About gpt-oss-20b
gpt-oss-20b is an AI model developed by OpenAI, in the language category, released in 2025, made available as an open-weights model with 20.9B params.
On this page you'll find gpt-oss-20b's full specifications, including 39 benchmark results. Review provider pricing and benchmark scores below, or compare gpt-oss-20b head-to-head with other language models.
Links
Run gpt-oss-20b locally — what to buy
Share this build ↗Sized to what gpt-oss-20b actually needs (~8k context at Q4), not to the biggest GPU: Good = cheapest that runs it, Better = best value, Best = most headroom. Speed is a hardware estimate; anything past ~120 tok/s is shown as instant.
Software support
✓ measured · · compatible · — not supported. Informational only — speed is hardware-based.
Community
No reports yet. Own this model on some hardware? Be the first to confirm it runs.
No comments yet. Start the discussion.
gpt-oss-20b benchmark scores▶
| Benchmark | Category | Score | Variant |
|---|---|---|---|
| TAU-bench Airline | agentic | 38% | high reasoning |
| TAU-bench Airline | agentic | 38% | high |
| TAU-bench Retail | agentic | 54.8% | high |
| TAU-bench Retail | agentic | 54.8% | high reasoning |
| Aider Polyglot | code | 34.2% | high |
| Aider Polyglot | code | 34.2% | high reasoning |
| Codeforces Elo | code | 2,516 | high-with-tools-elo |
| Codeforces Elo | code | 2,230 | high-no-tools-elo |
| Codeforces Elo | code | 2,230 | high reasoning, no tools |
| LiveCodeBench v5 | code | 73.22% | |
| SWE-bench Verified | code | 60.7% | high |
| SWE-bench Verified | code | 60.7% | high reasoning |
| SciCode | code | 37.9% | sub-problems |
| MMLU | general | 85.3% | high reasoning |
| MMLU | general | 85.3% | high |
| MMMLU | general | 75.7% | high-avg-14lang |
| MMMLU | general | 75.7% | high reasoning |
| AIME 2024 | math | 96% | high-with-tools |
| AIME 2024 | math | 92.1% | high-no-tools |
| AIME 2024 | math | 92.1% | high reasoning, no tools |
| AIME 2024 | math | 76.88% | reported in K2-Think |
| AIME 2025 | math | 98.7% | high-with-tools |
| AIME 2025 | math | 91.7% | high reasoning, no tools |
| AIME 2025 | math | 91.7% | high-no-tools |
| AIME 2025 | math | 74.58% | |
| HMMT February 2025 | math | 69.38% | |
| Omni-MATH-HARD | math | 41.51% | |
| GPQA Diamond | reasoning | 74.2% | high-with-tools |
| GPQA Diamond | reasoning | 71.5% | high reasoning, no tools |
| GPQA Diamond | reasoning | 71.5% | high-no-tools |
| GPQA Diamond | reasoning | 65.45% | |
| Humanity's Last Exam | reasoning | 17.3% | high-with-tools |
| Humanity's Last Exam | reasoning | 11.23% | |
| Humanity's Last Exam | reasoning | 10.9% | high reasoning, no tools |
| Humanity's Last Exam | reasoning | 10.9% | high-no-tools |
| HealthBench | safety | 42.5% | high reasoning |
| HealthBench | safety | 42.5% | high |
| HealthBench Hard | safety | 10.8% | high |
| HealthBench Hard | safety | 10.8% | high reasoning |
More models like gpt-oss-20b
Frequently asked questions
What is gpt-oss-20b? ▶
gpt-oss-20b is an AI model developed by OpenAI, in the language category, released in 2025. It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.
Who created gpt-oss-20b? ▶
gpt-oss-20b was developed by OpenAI and released in 2025.
Is gpt-oss-20b open source or proprietary? ▶
gpt-oss-20b ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.
How much does gpt-oss-20b cost? ▶
Pricing for gpt-oss-20b depends on the provider. See the providers table on this page for the latest API rates.
How does gpt-oss-20b perform on benchmarks? ▶
gpt-oss-20b is benchmarked across 39 evaluations on GenAIList, including TAU-bench Airline (38%). See the full benchmark table below and compare it with other models.
No videos yet. Know a good one? Add it below.
Know a great video about gpt-oss-20b?
Add YouTube Video
Add Tweet/Discussion
Report an Issue
Help us maintain accurate information in our database.
Suggest an Edit
Tell us what needs to be changed or corrected for