LLM
GA
Open weights
gpt-oss-120b
Strong open-weight reasoning on a single GPU
Release2025-08-05
Context131K
Max output32,768 tokens
Parameters117B
Local / VRAM48–80 GB
LicenseOpen weights
Overview
gpt-oss-120b is a model in the LLM category from OpenAI. It was released in 2025-08-05. The vendor is based in US. Best for: Strong open-weight reasoning on a single GPU.
Hardware: MXFP4 ~63 GB; fits on a single 80 GB GPU.
Strengths
- Apache-2.0 commercial use
- Single 80 GB GPU
- Strong tool use
Limits
- Weakens at very large context
- No image input
Specifications
| Vendor | OpenAI |
|---|---|
| Category | LLM |
| Release | 2025-08-05 |
| Status | GA |
| License | Open weights |
| License name | Apache-2.0 |
| Commercial use | Allowed |
| Parameters | 117B |
| Active parameters | 5.1B |
| Architecture | MoE transformer |
| Context window | 131K |
| Max output tokens | 32,768 |
| Input modalities | Text |
| Output modalities | Text |
| Prompt types | Text |
| Negative prompts | No |
| Max outputs | 1 |
| Local deployment | Yes |
| VRAM | 48–80 GB |
| System RAM | 64–128 GB |
| API access | Yes |
| Pricing | Free weights; hosting cost varies |
| Confidence | High |
| Vendor country | US |
| Vendor type | Frontier lab |
Pricing & hardware
Free weights; hosting cost varies
This model can be self-hosted or run locally. Approximate GPU memory: 48–80 GB. Approximate system RAM: 64–128 GB. API access is available.