gpt-oss-120b
gpt-oss-120b is OpenAI's larger open-weight reasoning model, released in August 2025 under Apache 2.0. It is a 128-expert mixture-of-experts with 117B total and 5.1B active parameters, natively quantized to MXFP4 so it fits on a single 80 GB GPU. Reasoning effort can be set to low, medium or high.
Specifications
- Context window
- 128Kabout 98,000 words, or 200 pages
- Max output
- Not published
- Input price
- $0.15per 1M tokens
- Output price
- $0.60per 1M tokens
- Cached input
- $0.015per 1M tokens
- Weights
- OpenApache-2.0
- Parameters
- 117B total, 5.1B activeMoE
- Input types
- Text
- Output types
- Text
- Reasoning mode
- Yes
- Tool calling
- Yes
- Released
- Aug 5, 2025Current
Fireworks standard serverless price; Groq lists $0.15 / $0.60 with up to 65,536 completion tokens.
What it costs in practice
| Workload | Monthly cost at list price |
|---|---|
| Support chatbot30,000 conversations a month, 2,000 input and 400 output tokens each | $16 |
| Document summaries2,000 documents a month, 12,000 input and 800 output tokens each | $4.56 |
| Coding agent1,000 tasks a month, 60,000 input and 6,000 output tokens each | $13 |
Standard rates, no caching or batch discounts. Run your own numbers.
Strengths
- Fits on one 80 GB GPU thanks to MXFP4 weights
- Adjustable reasoning effort with browsing, Python and function-calling formats
- Apache 2.0 and very widely hosted
Trade-offs to weigh
- Text-only input
- 128K context is modest by 2026 standards
Where it fits
OpenAI positions gpt-oss-120b for:
Where you can use it
- Hugging Face
- Ollama
- Groq
- Fireworks
- Together AI
- OpenRouter
- AWS
- Azure
Run it yourself
Released under Apache-2.0. Approximate memory needed just to load the weights:
| Precision | Memory for weights |
|---|---|
| 16-bit (BF16/FP16) | 234 GB |
| 8-bit (FP8/INT8) | 117 GB |
| 4-bit (Q4_K_M, about 4.85 bits) | 71 GB |
This is a mixture-of-experts model: all 117B total parameters must sit in memory, but only about 5.1B are used per token, so it runs faster than a dense model of the same size.
Add memory for the KV cache (it grows with context length and concurrent users) and 1-2 GB of runtime overhead. Size it precisely in the VRAM calculator.
Alternatives to consider
| Model | Tier | Context | Input / output per 1M | |
|---|---|---|---|---|
| DeepSeek-V4.1-FlashDeepSeek | Advanced | 1M | $0.30 / $1.20 | Compare |
| MiniMax-M3MiniMax | Advanced | 1M | $0.30 / $1.20 | Compare |
| Mistral Large 3Mistral AI | Advanced | 256K | $0.50 / $1.50 | Compare |
| NVIDIA Nemotron 3 UltraNVIDIA | Advanced | 1M | $0.50 / $2.20 | Compare |
Frequently asked questions
How much does gpt-oss-120b cost?
gpt-oss-120b lists at $0.15 per million input tokens and $0.60 per million output tokens, with cached input at $0.015 (checked October 10, 2026). Fireworks standard serverless price; Groq lists $0.15 / $0.60 with up to 65,536 completion tokens. Confirm current rates on the provider's pricing page before you commit.
What is the context window of gpt-oss-120b?
gpt-oss-120b accepts up to 131,072 tokens of context (about 98,000 words, or 200 pages). Long prompts cost more and can be slower, so most apps send only what the task needs.
Is gpt-oss-120b open source?
gpt-oss-120b is an open-weight model released under the Apache-2.0 license. You can download the weights from Hugging Face (openai/gpt-oss-120b) and run or fine-tune them yourself, within the license terms. Open weights isn't the same as fully open source: training data and code may not be published.
Can I run gpt-oss-120b locally?
Yes, if you have enough memory. The weights alone need about 71 GB at 4-bit quantization and about 234 GB at 16-bit, plus room for the KV cache. Use the VRAM calculator on LLModel.com to check a specific GPU or Mac.
What is gpt-oss-120b best for?
OpenAI positions gpt-oss-120b for reasoning, agents, self-hosting, high volume. It accepts text as input, has a reasoning (thinking) mode and supports tool calling. Test it on a sample of your own tasks before you choose it for production.
Sources
- openai.com/index/introducing-gpt-oss
- docs.fireworks.ai/serverless/pricing
- openai.com/open-models
- huggingface.co/openai/gpt-oss-120b
- huggingface.co/openai/gpt-oss-120b/raw/main/config.json
- console.groq.com/docs/models
- ollama.com/library
Facts checked against these pages on October 10, 2026. Spot an error? Tell us.
Still choosing a model?
Send us your use case and constraints. We'll come back with a shortlist and an evaluation plan you can run on your own data.
- Free and vendor-neutral: no model provider pays for placement.
- Your details go only to partners you agree to.