gpt-oss-120b

gpt-oss-120b by openai: a 116.8B-parameter mixture-of-experts open-weight model under the apache-2.0 license.

apache-2.0Mixture of experts131K contextAuto-imported · not yet reviewed

5,117,998 downloads · 5,268 likes · Model card · synced 2026-09-19

Specifications

Parameters: 116.8B (3.7B active)
Architecture: moe · gpt_oss
Layers: 36
Hidden size: 2,880
KV heads · head dim: 8 · 64
Vocabulary: 201,088
Native dtype: —
Context window: 131,072 tokens
Released: 2025-08-04
Last modified on Hub: 2025-08-26

Features & licensing

License: apache-2.0
Commercial use: Yes
Modalities: text
Tool / function calling: No
Reasoning mode: No
Languages: EN
Quantised variants on Hub: GGUF, bitsandbytes
Pipeline: text-generation

transformerssafetensorsgpt_osstext-generationvllmconversationaleval-resultsendpoints_compatible8-bitmxfp4

Task fit

Editorial scores (0–100) used by the recommendation engine.

Chat70
RAG / Q&A70
Code70
Summarisation70
Extraction70
Agents70

Hardware to run gpt-oss-120b

Weights need about 257 GB at FP16, 135 GB at INT8 and 77 GB at INT4. The KV cache adds roughly 74 MB per 1,000 tokens per request. For a reference workload of 5 requests per second with 1,500 input and 300 output tokens, total GPU memory is around 101.7 GB, which fits on 2 × NVIDIA A100 (80 GB).

GPU configurations at the reference workload
Configuration	VRAM	Utilisation	Est. first token	Cloud / month
5 × NVIDIA L4 (24 GB)	120 GB	85 %	~0.5 s	$2,920
3 × NVIDIA L40S (48 GB)	144 GB	71 %	~0.2 s	$4,161
2 × NVIDIA A100 (80 GB)recommended	160 GB	64 %	~0.1 s	$4,672
2 × NVIDIA H100 (80 GB)	160 GB	64 %	~0.0 s	$6,570

Estimates only. Run an assessment for your own traffic.

Auto-imported from Hugging Face. Architecture, license and feature data are synced from the model repository; the task-fit scores below are provisional defaults until a curator reviews them.