ModelBeat
Models/gpt-oss-120b
All models
O
openai/gpt-oss-120b

gpt-oss-120b is OpenAI’s open-weights AI model, released Aug 5, 2025. It costs $0.04/M input and $0.17/M output tokens, and handles a 131K-token context window.

gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases.

Context
131K
Input
$0.04 / 1M
Output
$0.17 / 1M
Released
Aug 5, 2025
Modalities
texttext

Benchmarks

Scores on standardized evaluations. Higher is better — and rank shows where gpt-oss-120b lands among all models tracked on Model Beat.

31.0
Intelligence Index
Epoch AI
31st percentile of tracked models
43.0
Coding Index
Epoch AI
43rd percentile of tracked models
10.0
Agentic Index
Epoch AI
10th percentile of tracked models

Reasoning

6 evals
GPQA Diamond75.8%

Graduate-level scientific reasoning

Humanity's Last Exam18.5%

Frontier of human expert knowledge

MMLU-Pro80.8%

Broad expert knowledge

SimpleBench22.1%

Common-sense trick questions

SimpleQA Verified13.9%

Factual accuracy & hallucination

WeirdML48.2%

Novel ML problem-solving

Coding

2 evals
LiveCodeBench87.8%

Contamination-free coding

SciCode38.9%

Scientific research coding

Math

1 evals
AIME 2024/202588.9%

Olympiad-qualifier math

Agentic & Tools

4 evals
APEX4.7%

Multi-step agentic tasks

METR task horizon42 min

Autonomous task length

Terminal-Bench18.7%

Command-line agentic tasks

τ²-bench65.8%

Tool-agent-user reliability

In the news

2

Changelog

5
  • gpt-oss-120b: cheapest credible provider via DeepInfra now $0.037 per 1M input tokens (was $0.03, +23%)cheapest provider · Jul 25, 2026
  • gpt-oss-120b: cheapest credible provider via DeepInfra now $0.17 per 1M output tokens (was $0.18, -6%)cheapest provider · Jul 25, 2026
  • gpt-oss-120b: cheapest credible provider via DekaLLM now $0.18 per 1M output tokens (was $0.15, +20%)cheapest provider · Jul 18, 2026
  • gpt-oss-120b: cheapest credible provider via OpenInference now $0.03 per 1M input tokens (was $0.036, -17%)cheapest provider · Jul 15, 2026
  • gpt-oss-120b: cheapest credible provider via OpenInference now $0.15 per 1M output tokens (was $0.18, -17%)cheapest provider · Jul 15, 2026

Compare gpt-oss-120b with

Frequently asked questions

What is gpt-oss-120b?

gpt-oss-120b is an open-weight, 117B-parameter Mixture-of-Experts (MoE) language model from OpenAI designed for high-reasoning, agentic, and general-purpose production use cases.

Who created gpt-oss-120b?

gpt-oss-120b is an AI model developed by OpenAI, part of the GPT family, released Aug 5, 2025.

How much does gpt-oss-120b cost?

gpt-oss-120b is served from $0.04 per million input tokens and $0.17 per million output tokens (cheapest credible provider on OpenRouter: DeepInfra).

What is gpt-oss-120b's context window?

gpt-oss-120b supports a context window of up to 131K tokens.

How does gpt-oss-120b perform on benchmarks?

On standardized evaluations tracked by Epoch AI, gpt-oss-120b scores 87.8% on LiveCodeBench, 88.9% on AIME 2024/2025, 48.2% on WeirdML.

Is gpt-oss-120b open source?

gpt-oss-120b is released with open weights (Open weights (unrestricted)), so the model can be downloaded and self-hosted.

Benchmark data from Epoch AI (CC BY) and Artificial Analysis; pricing, specs & descriptions from OpenRouter. Data updated Jul 30, 2026.