gpt-oss-120b
OpenAI · released
OpenAI's largest open-weights model: a text-only reasoning model from August 2025 that can be self-hosted on a single H100-class GPU and fine-tuned. It trails OpenAI's API models by a wide margin, and no newer open-weights release has followed.
Best for: Self-hosting · Fine-tuning · Private deployments
- Intelligence Index
- 11.6
- #37 of 38 · at High effort
- Price per 1M tokens
- $0.15 in $0.59 out
- Blended $0.26 at 3:1
- Context window
- 131K
- Up to 131K output
- Output speed
- 183 tokens/s
- $0.11 measured cost per task
Capability profile
One 0–100 score per area, placing this model between the weakest and strongest tracked model across that area's benchmarks.
- Overall3 · #37 of 38
Based on 2 of 3 benchmarks
- Coding0 · #39 of 39
Based on 2 of 5 benchmarks
- Agents & knowledge work8 · #37 of 40
Based on 1 of 6 benchmarks
- Reasoning & science11 · #34 of 39
Based on 2 of 2 benchmarks
- Knowledge & reliability6 · #39 of 39
Based on 2 of 2 benchmarks
- VisionNot enough data
Based on 0 of 2 benchmarks
Benchmark results
All third-party results on record. Rank is among the tracked models that have been tested on that benchmark.
Overall
#37 of 38· High effort· Artificial Analysis
- Vals IndexNot tested
- Arena · Text1352 ±4
#32 of 33· 30,018 votes· Arena
Coding
#35 of 38· High effort· Artificial Analysis
- SciCode34.0%
#38 of 38· High effort· Artificial Analysis
- Arena · WebDevNot tested
- Vibe Code BenchNot tested
- Code MigrationNot tested
Agents & knowledge work
- GDPval-AA5.6
#36 of 39· High effort· Artificial Analysis
- Arena · AgentNot tested
- Finance Agent v2Not tested
- Excel Modeling BenchmarkNot tested
- Legal Research BenchNot tested
- Tax Agent BenchNot tested
Reasoning & science
- Humanity's Last Exam19.6%
#34 of 39· High effort· Artificial Analysis
- CritPt1.1%
#34 of 39· High effort· Artificial Analysis
- ARC-AGI-2saturatingNot tested
- ARC-AGI-3Not tested
- GPQA Diamondsaturating78.2%
#24 of 27· High effort· Artificial Analysis
Knowledge & reliability
#38 of 39· High effort· Artificial Analysis
- AA-LCR52.0%
#38 of 39· High effort· Artificial Analysis
Vision
- MMMU-ProNot tested
- Arena · VisionNot tested
Reported by OpenAI
Numbers the lab published at launch, run on its own harness and settings. Useful context, but not independently verified and never used in the rankings on this site.
- GPQA Diamond80.1%
- Humanity's Last Exam19%
Specifications
- Lab
- OpenAI
- Released
- Aug 5, 2025
- Availability
- Generally available
- Context window
- 131,072 tokens
- Max output
- 131,072 tokens
- Accepts
- Text
- Knowledge cutoff
- Jun 2024
- Weights
- Open
- Licence
- Apache 2.0
- Parameters
- 117B total · 5.1B active
- Reasoning control
- Effort low to high
- API model ID
gpt-oss-120b
Pricing
- Input / 1M tokens
- $0.15
- Cached input / 1M
- $0.13
- Output / 1M tokens
- $0.59
- Blended (3:1)
- $0.26
- Measured cost / task
- $0.11
Typical price at third-party hosts. The lab does not sell paid API access to this model.
Official sources
Closest alternatives
The models nearest to this one on overall score.
23 benchmarks tracked. See methodology and sources.