v2.2 · 2026-06-12

GPT-OSS 120B

gpt-oss-120b

REASONING

GPT-OSS 120B is a 117B reasoning model profile for local AI planning. This page records provider, license, context, quantization, hardware-fit estimates, setup hints, source links, and caveats so readers do not choose a model by name alone.

PROVIDER

OpenAI

FAMILY

GPT-OSS

MODEL TYPE

reasoning

PARAMETERS

117B (5.1B active)

MODALITIES

text

ARCHITECTURE

MoE Transformer; alternating dense and locally banded sparse attention; grouped multi-query attention; 36 layers; 128 experts; 4 active experts per token; MXFP4 MoE weights

CONTEXT WINDOW

131,072 tokens

TRAINING TOKENS

—

RELEASE DATE

2025-08-05

LICENSE & LINKS

Apache 2.0license text

MODEL CARD

huggingface.co/openai/gpt-oss-120b

HUGGING FACE

openai/gpt-oss-120b

GGUF REPOSITORIES

lmstudio-community/gpt-oss-120b-GGUF community conversion

Community GGUF repository referenced by the June 2026 model research.

bartowski/openai_gpt-oss-120b-GGUF community conversion

Community GGUF repository referenced by the June 2026 model research.

SETUP HINTS

OLLAMA

ollama run gpt-oss:120b

LM STUDIO

Available according to the supplied June 2026 research; select a cited GGUF/MLX build and verify current app metadata before relying on hands-on setup guidance.

LLAMA.CPP

Use the cited GGUF repository (lmstudio-community/gpt-oss-120b-GGUF) with current llama.cpp-compatible tooling.

QUANTIZATION FILE SIZES (GGUF)

Q4_K_M

62.84 GB

Q8_0

63.4 GB

Quantization note: Q4_K_M for first local attempts unless a cited runtime recommends a different default.

RAM / VRAM ESTIMATES

MIN RAM

65 GB

COMFORTABLE RAM

80 GB

MIN VRAM

65 GB

COMFORTABLE VRAM

80 GB

These are conservative local-inference estimates, not official hardware requirements.

HARDWARE FIT

CPU only (no GPU)

8 GB RAM

16 GB RAM

32 GB RAM

8 GB VRAM

12 GB VRAM

24 GB VRAM

Apple Silicon (unified memory)

Hardware fit values are conservative local-inference estimates based on GGUF size plus runtime and KV-cache overhead. Actual requirements depend on context length, quantization, runtime, GPU offload, and other running apps.

BEST FOR

·Server-side local reasoning tests, agents, and teams with 60GB-plus memory available for experimentation and deployment.

AVOID IF

·Avoid if you lack workstation-class hardware or want a simple beginner laptop model to run.

CAVEATS

·Consumer-hostile for local beginners; realistic local use requires workstation, server, or very high-memory unified-memory hardware.

·Place this in an advanced or server-only section, not the main beginner catalog.

·OpenAI positions it as fitting a single 80GB GPU; local guides describe it as best with 60GB-plus VRAM or unified memory.

·The common Ollama and LM Studio packages are about 65GB, before considering runtime overhead.

·Tool use depends on the app or server exposing tools; the model alone does not provide web browsing, Python, or function execution.

·Higher reasoning effort can be useful but will materially affect latency.

·The 128k context window is an upper limit; full-context use can be slow and memory-heavy.

·Common third-party quants are not much smaller than MXFP4; Q5 size was not confirmed in cited sources in the checked 120B GGUF source.

·Production use needs separate safeguards because open weights can be modified, fine-tuned, or deployed without OpenAI-side mitigations.

·Commercial-use status from the supplied research: permitted, subject to OpenAI gpt-oss usage policy

·Beginner summary from supplied research: GPT-OSS 120B is the large open-weight reasoning model, aimed at serious reasoning, tools, and production experiments. It is not beginner-friendly locally because realistic use requires server, workstation, or high-memory unified hardware rather than consumer laptops.

SOURCE URLS