O
gpt-oss-120bOPEN
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.
Prices update automatically — checked daily against provider list prices.
- $0.07 per 1M tokens, assuming 3 input for 1 output the convention used by Artificial Analysis
- cheaper than 83% of the 191 open-weight models tracked here
- 3 of 88 OpenAI models tracked here, from cheapest
Benchmarks & performance
Independent benchmark scores for gpt-oss-120b, measured by Artificial Analysis. Higher is better. Measurement mode: high.
For the same budget, gpt-oss-120b delivers 15 times more capability. Open weights also mean you can self-host it and pay nothing per token.
Models at this level cost $0.96 per 1M tokens (median of 20) — this one costs $0.07.
Cheaper and better on this index: DeepSeek V4 Flash 0731 · Solar Pro 4 · Ling-3.0-flash
About this model
gpt-oss-120b is an open-weight AI model by OpenAI. You can download and self-host it for free; the prices below are hosted-API list prices, tracked daily, for when you prefer convenience over self-hosting.
Frequently asked questions
What is gpt-oss-120b?
gpt-oss-120b is an AI language model from OpenAI. It is open-weight: you can download it and run it on your own hardware, for free. It scores 24.1 on the Artificial Analysis intelligence index.
Is gpt-oss-120b free?
The weights are free and open — you can self-host gpt-oss-120b and pay nothing per token. If you prefer a hosted API, list prices are $0.04 per million input tokens and $0.17 per million output tokens.
What is gpt-oss-120b good at?
Independent benchmarks from Artificial Analysis give it GPQA 78.2%, MMLU-Pro 80.8%, Humanity's Last Exam 19.6%, Long Context Reasoning 51%, LiveCodeBench 87.8%, SciCode 38.9%, AIME 2025 93.4%, IFBench 69%, τ²-Bench 65.8%, τ-Bench Banking 12.8%, Terminal-Bench 26.2%, Terminal-Bench Hard 23.5%. It is particularly used for code generation.
How fast is gpt-oss-120b?
It generates about 167.3 tokens per second, with a median 0.53s delay before the first token. Measured independently by Artificial Analysis.
Can I self-host gpt-oss-120b?
Yes. gpt-oss-120b has open weights, so you can download it and run it on your own GPU or server with tools like Ollama, vLLM or llama.cpp — with no per-token cost.