Hugging FaceOpenAI's open-weight reasoning model, designed for reasoning problems and agent tasks.
- Rating
- 3.5
- Downloads
- 366K
- Parameters
- 117B, 5.1B active
- Context
- 128K
- Download
- 63 to 65 GB
- Commercial use
- Yes
Too big for this Mac
Its smallest size needs 65 GB of memory, more than M4 · 16 GB can give it.
About
Reasoning effort can be set to low, medium or high, to match the task and how long you can wait. Its full chain of thought is open to read, which helps when checking how it reached an answer. It can be further trained to suit one specific use.
How it compares
In LM Arena's public vote for text, it places 36th of the 62 open models rated.
In LM Arena's public vote for coding, it places 39th of the 59 open models rated.
Limitations
OpenAI says the raw chain of thought is not meant to be shown to end users.
Good to know
OpenAI released the code to run it and its tool setups along with the weights.
Its sibling gpt-oss-20b was made for lower wait times and local use.
Sources: openai/gpt-oss-120b model card, gpt-oss-120b & gpt-oss-20b Model Card, gpt-oss-120b & gpt-oss-20b Model Card, Outmake's speed estimates, LM Arena and Hugging Face. Every line comes from one of these.
Ratings
3.5 out of 5, from how people voted on its answers.
LM Arena ratings of October 28, 2026 (CC BY 4.0): people vote between two answers without knowing which model wrote which. The bar ends at 1488, the best any open model has. The Arena tests the full model; a small size on a Mac answers a little worse.
Good at
Licence
Apache 2.0. Commercial use: yes. The maker's page has the full terms.
GPT-OSS 120B runs well from a Mac with an M2 Max and 96 GB of memory, at UD-Q8_K_XL, writing about 29 words (39 tokens) a second. 15 of 59 Mac configurations run it well.
Compare with
The nearest models in size, judged for the same Mac.
- Qwen3.5 122BFor long documents and questions in many languages, and it can read photos and screenshots too.Too big for this Mac · 58 GBGet
- Qwen3.8 Flash NextPreview of Qwen's next designToo big for this Mac · 91 GBGet
- MiniMax M2.7Help finding bugs in codeToo big for this Mac · 102 GBGet
- DeepSeek V4 FlashLong documents and tool useToo big for this Mac · 128 GBGet
Sizes
| Size | Download | Memory at 8K | Memory at 32K | Open |
|---|---|---|---|---|
| UD-Q4_K_XLBalanced | 63 GB | 63 GB | 63 GB | Open |
| UD-Q8_K_XLBest | 64 GB | 64 GB | 64 GB | Open |
All 16 sizes
| Size | Download | Memory at 8K | Memory at 32K | Open |
|---|---|---|---|---|
| Q3_K_S | 63 GB | 63 GBestimated | 63 GB | Open |
| Q2_K | 63 GB | 63 GBestimated | 63 GB | Open |
| Q4_0 | 63 GB | 63 GBestimated | 63 GB | Open |
| Q3_K_M | 63 GB | 63 GBestimated | 63 GB | Open |
| Q4_1 | 63 GB | 63 GBestimated | 63 GB | Open |
| Q4_K_S | 63 GB | 63 GBestimated | 63 GB | Open |
| Q4_K_M | 63 GB | 63 GBestimated | 63 GB | Open |
| Q2_K_L | 63 GB | 63 GBestimated | 63 GB | Open |
| Q5_K_S | 63 GB | 63 GBestimated | 63 GB | Open |
| Q5_K_M | 63 GB | 63 GBestimated | 63 GB | Open |
| Q6_K | 63 GB | 63 GBestimated | 64 GB | Open |
| UD-Q6_K_XL | 63 GB | 63 GBestimated | 64 GB | Open |
| Q8_0 | 63 GB | 63 GBestimated | 64 GB | Open |
| F16 | 65 GB | 65 GBestimated | 65 GB | Open |
Memory is what llama.cpp's own calculator reported for each size when the catalog was built, with the app's settings. Smaller, Balanced and Best are the three sizes the store names.
Run It Anywhere
GPT-OSS 120B is an open model, so any app that runs GGUF files can run it. These take the UD-Q4_K_XL size from unsloth straight from Hugging Face.
- LM StudioMac, Windows, Linux
Open - JanMac, Windows, Linux
Open - MstyMac, Windows, Linux
Open - OllamaMac, Windows, Linux
ollama run hf.co/unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL - llama.cppMac, Windows, Linux
llama-server -hf unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL - Docker Model RunnerMac, Windows, Linux
docker model run hf.co/unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL
On a PC Graphics Card
| Video memory | Largest size that fits | Download | Memory at 8K |
|---|---|---|---|
| 6 GB | None worth running | needs 63 GB | |
| 8 GB | None worth running | needs 63 GB | |
| 10 GB | None worth running | needs 63 GB | |
| 11 GB | None worth running | needs 63 GB | |
| 12 GB | None worth running | needs 63 GB | |
| 16 GB | None worth running | needs 63 GB | |
| 20 GB | None worth running | needs 63 GB | |
| 24 GB | None worth running | needs 63 GB | |
| 32 GB | None worth running | needs 63 GB | |
| 48 GB | None worth running | needs 63 GB |
The largest size, of those that keep three bits a weight or more, whose memory at an 8K chat is under 92% of the card's; the rest is the desktop's. Estimated from the files, the same way as for a Mac; speed on a PC is not estimated yet. A size that doesn't fit the card still runs, slower, with part of it in ordinary memory.
On every Mac
| Mac | Size | Fit | Memory | Writes | Reads |
|---|---|---|---|---|---|
| M6 · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M6 · 24 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M6 · 32 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 · 24 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 · 32 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 Pro · 24 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 Pro · 48 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 Pro · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 Max · 36 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 Max · 48 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 Max · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M5 Max · 128 GB | UD-Q8_K_XL | Runs great | 66 GB | 53 | 1593 |
| M5 Ultra · 96 GB | UD-Q8_K_XL | Runs well | 66 GB | 91 | 2549 |
| M5 Ultra · 256 GB | UD-Q8_K_XL | Runs great | 66 GB | 91 | 3187 |
| M5 Ultra · 512 GB | UD-Q8_K_XL | Runs great | 66 GB | 91 | 3187 |
| M4 · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 · 24 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 · 32 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 Pro · 24 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 Pro · 48 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 Pro · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 Max · 36 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 Max · 48 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 Max · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M4 Max · 96 GB | UD-Q8_K_XL | Runs well | 66 GB | 28 | 428 |
| M4 Max · 128 GB | UD-Q8_K_XL | Runs great | 66 GB | 34 | 535 |
| M3 · 8 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 · 24 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 Pro · 18 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 Pro · 36 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 Max · 36 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 Max · 48 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 Max · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M3 Max · 96 GB | UD-Q8_K_XL | Runs well | 66 GB | 24 | 332 |
| M3 Max · 128 GB | UD-Q8_K_XL | Runs great | 66 GB | 30 | 443 |
| M3 Ultra · 96 GB | UD-Q8_K_XL | Runs well | 66 GB | 45 | 665 |
| M3 Ultra · 256 GB | UD-Q8_K_XL | Runs great | 66 GB | 45 | 665 |
| M3 Ultra · 512 GB | UD-Q8_K_XL | Runs great | 66 GB | 45 | 886 |
| M2 · 8 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 · 24 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 Pro · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 Pro · 32 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 Max · 32 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 Max · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 Max · 96 GB | UD-Q8_K_XL | Runs well | 66 GB | 29 | 413 |
| M2 Ultra · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M2 Ultra · 128 GB | UD-Q8_K_XL | Runs great | 66 GB | 45 | 653 |
| M2 Ultra · 192 GB | UD-Q8_K_XL | Runs great | 66 GB | 45 | 653 |
| M1 · 8 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M1 · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M1 Pro · 16 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M1 Pro · 32 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M1 Max · 32 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M1 Max · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M1 Ultra · 64 GB | UD-Q4_K_XL | Too big | 65 GB | ||
| M1 Ultra · 128 GB | UD-Q8_K_XL | Runs great | 66 GB | 34 | 484 |
Writes and reads are words a second (a token is about three quarters of a word), 4K tokens into a chat, estimated unless marked measured. The size is the one the store offers that Mac. Constants fitted October 11, 2026; how, and how wrong.
Badge for the model card
Shows on GPT-OSS 120B's Hugging Face card which Macs run it well, linked to this page. It updates every morning.
Sizes and files are read from unsloth/gpt-oss-120b-GGUF on Hugging Face and checked every morning. The rating is the model's score on the LM Arena text leaderboard (CC BY 4.0), where people vote between two answers without knowing which model wrote which. Catalog of November 4, 2026.