RUN THIS LLM
Search local LLM hardware requirements
GPT-OSS 120B
OpenAI · 117B MoE · General
OpenAI's first open-weight model. 117B MoE with only 5.1B active params. Apache 2.0. RL-trained reasoning with configurable thinking depth. 128K context.
VRAM Requirements
| Quantization | VRAM |
|---|---|
| Q4_K_M (smallest) | 70.2 GB |
| Q8_0 (balanced) | 128.7 GB |
| FP16 (full quality) | 234 GB |
Specifications
- Parameters: 117B MoE (5.1B active per token)
- Category: General
- Max context: 128K tokens
- System RAM: 96 GB minimum
- HuggingFace: openai/gpt-oss-120b
Benchmarks
- MMLU: 86 — General knowledge & reasoning (5-shot)
- HumanEval: 88 — Code generation (pass@1)
- MATH: 74 — Competition-level math reasoning
Loading interactive analysis...