// gpt-oss · Text LLM
gpt-oss-120b: VRAM & local deployment
117B total · 5.1B active (MoE) parameters. About 78.8 GB at 4-bit and 258 GB at BF16 with one 8K-token request.
released 2025-08-04licence: Commercial OKmemory figures are estimates
01Specs
| Publisher | OpenAI · openai |
|---|---|
| Family / series | gpt-oss · gpt-oss |
| Type | Text LLM (text-generation) |
| Parameters | 117B total · 5.1B active (MoE) Source: official model card“117B parameters with 5.1B active parameters” |
| Context length | 128K tokens (131,072)max_position_embeddings in config.json |
| Architecture | 36 layers (18 full-attention) · 18 sliding-window layers (128) · GQA · MoE |
| Released weights | packed quantised tensors (U8, e.g. MXFP4 / INT4) · 65.3 GBdownload |
| Licence | Apache-2.0 · Commercial OKPermissive licence (Apache-2.0 / MIT / CC-BY / OpenMDW): commercial use allowed. |
| Released | 2025-08-04 (Hugging Face repo created) |
02How much VRAM it needs (estimate)
Total = weights + 10% + 0.5GB overhead + KV cache for one request (FP16 KV). More concurrent users multiply the KV column.
| Precision | Weights | Total @ 8K | Total @ 32K | Total @ 128K | Smallest GPU | Mac |
|---|---|---|---|---|---|---|
| BF16 / FP16 (original) | 234 GB | 258 GB | 259 GB | 263 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB |
| FP8 / INT8 / Q8_0 | 124 GB | 138 GB | 138 GB | 142 GB | 4 × RTX PRO 6000 (384GB) | Mac 256GB |
| 4-bit (Q4_K_M / AWQ / GPTQ-Int4) | 70.9 GB | 78.8 GB | 79.7 GB | 83.4 GB | RTX PRO 6000 96GB | Mac 128GB |
KV cache per request: 8K 0.3 GB · 32K 1.2 GB · 128K 4.8 GB
Bits per weight: BF16 16, FP8 8.5, 4-bit 4.85. “Fits” = total ≤ 90% of memory; Mac GPU can use ~2/3 of unified memory (≤36GB) or ~3/4 (larger). For MoE, all experts must be in memory, so size by total parameters. Method →
03How to run it
Only runtimes the publisher or the runtime’s official library names for this model.
- vLLM — mentioned in the official model card
- LM Studio — mentioned in the official model card
- Transformers — mentioned in the official model card
- Ollama — in the official Ollama library · gpt-oss:120b
ollama run gpt-oss:120b04Official repositories & quantised builds
1 official repo(s) from openai.
- gpt-oss-120b main
05Other gpt-oss models
All 4 →| Model | Params | 4-bit total | Released |
|---|---|---|---|
| gpt-oss-20b · same series | 21B / 3.6B | 14.7 GB | 2025-08-04 |
| gpt-oss-safeguard-20b | 21B / 3.6B | 14.7 GB | 2025-09-18 |
| gpt-oss-safeguard-120b | 117B / 5.1B | 78.8 GB | 2025-09-18 |
06Sources
Data checked 2025-08-26. Parameters, context and licence are copied from the official repository; memory figures are our estimates.