Hugging Face
A community version of another maker's model.
- Rating
- 2.8
- Downloads
- 101K
- Likes
- 402
- Parameters
- 33B
- Context
- 32K
- Licence
- apache-2.0
Runs well on M4 · 16 GB
The IQ2_XXS size from bartowski is the one to get: a 9.0 GB download that uses 10 GB of memory, and writes about 3.8 words a second.
Ratings
2.8 out of 5, across people's blind votes 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.
Qwen2.5 Coder 32B runs well from a Mac with an M1 and 16 GB of memory. 56 of 59 Mac configurations run it well. It is a re-tune or merge made by someone other than the original model's maker, so its ratings, where it has any, are its own.
Compare with
The nearest models in size.
- Qwen2.5 32BAlibaba Qwen · 833K downloadsRuns well · 9.0 GB · 3.8 words/s
- Qwen3 32BAlibaba Qwen · 245K downloads · Arena 1347Runs well · 9.3 GB · 3.8 words/s
- QwQ 32BAlibaba Qwen · 74K downloads · Arena 1336Runs well · 9.0 GB · 3.8 words/s
- QwQ 32B PreviewAlibaba Qwen · 6.9K downloads · Arena 1154Tight fit · 10 GB · 3.6 words/s
Sizes
From bartowski/Qwen2.5-Coder-32B-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.
| Size | Download | Memory at 8K | Get |
|---|---|---|---|
| IQ2_XXS | 9.0 GB | 10 GBestimated | Open |
| IQ2_XS | 10.0 GB | 11 GBestimated | Open |
| IQ2_S | 10 GB | 12 GBestimated | Open |
| IQ2_M | 11 GB | 13 GBestimated | Open |
| IQ3_XXS | 13 GB | 14 GBestimated | Open |
| Q2_K_L | 13 GB | 14 GBestimated | Open |
| IQ3_XS | 14 GB | 15 GBestimated | Open |
| Q3_K_S | 14 GB | 16 GBestimated | Open |
| IQ3_M | 15 GB | 16 GBestimated | Open |
| Q3_K_L | 17 GB | 19 GBestimated | Open |
| IQ4_XS | 18 GB | 19 GBestimated | Open |
| Q3_K_XL | 18 GB | 19 GBestimated | Open |
| Q4_0_4_4 | 19 GB | 20 GBestimated | Open |
| Q4_0_4_8 | 19 GB | 20 GBestimated | Open |
| Q4_0_8_8 | 19 GB | 20 GBestimated | Open |
| IQ4_NL | 19 GB | 20 GBestimated | Open |
| Q4_K_S | 19 GB | 20 GBestimated | Open |
| Q4_K_L | 20 GB | 22 GBestimated | Open |
| Q5_K_S | 23 GB | 24 GBestimated | Open |
| Q5_K_L | 24 GB | 25 GBestimated | Open |
| Q6_K_L | 27 GB | 29 GBestimated | Open |
Run It Anywhere
Qwen2.5 Coder 32B is an open model, so any app that runs GGUF files can run it. These take the IQ2_XXS size from bartowski straight from Hugging Face.
- LM StudioMac, Windows, Linux
Open - JanMac, Windows, Linux
Open - MstyMac, Windows, Linux
Open - OllamaMac, Windows, Linux
ollama run hf.co/bartowski/Qwen2.5-Coder-32B-GGUF:IQ2_XXS - llama.cppMac, Windows, Linux
llama-server -hf bartowski/Qwen2.5-Coder-32B-GGUF:IQ2_XXS - Docker Model RunnerMac, Windows, Linux
docker model run hf.co/bartowski/Qwen2.5-Coder-32B-GGUF:IQ2_XXS
On a PC Graphics Card
| Video memory | Largest size that fits | Download | Memory at 8K |
|---|---|---|---|
| 6 GB | None worth running | needs 14 GB | |
| 8 GB | None worth running | needs 14 GB | |
| 10 GB | None worth running | needs 14 GB | |
| 11 GB | None worth running | needs 14 GB | |
| 12 GB | None worth running | needs 14 GB | |
| 16 GB | Q3_K_S | 14 GB | 16 GB |
| 20 GB | Q3_K_XL | 18 GB | 19 GB |
| 24 GB | Q4_K_L | 20 GB | 22 GB |
| 32 GB | Q6_K_L | 27 GB | 29 GB |
| 48 GB | Q6_K_L | 27 GB | 29 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 |
|---|---|---|---|---|
| M6 · 16 GB | IQ2_XXS | Runs well | 10 GB | 5.5 |
| M6 · 24 GB | Q3_K_S | Runs well | 16 GB | 4.2 |
| M6 · 32 GB | Q4_K_M | Runs well | 21 GB | 3.4 |
| M5 · 16 GB | IQ2_XXS | Runs well | 10 GB | 5 |
| M5 · 24 GB | Q3_K_S | Runs well | 16 GB | 3.8 |
| M5 · 32 GB | Q4_K_M | Runs well | 21 GB | 3.1 |
| M5 Pro · 24 GB | Q3_K_S | Runs well | 16 GB | 7.5 |
| M5 Pro · 48 GB | Q4_K_L | Runs great | 23 GB | 6 |
| M5 Pro · 64 GB | Q6_K_L | Runs great | 30 GB | 5.1 |
| M5 Max · 36 GB | Q4_K_L | Runs well | 23 GB | 8.9 |
| M5 Max · 48 GB | Q4_K_L | Runs great | 23 GB | 12 |
| M5 Max · 64 GB | Q6_K_L | Runs great | 30 GB | 9.9 |
| M5 Max · 128 GB | Q8_0 | Runs great | 40 GB | 9.4 |
| M5 Ultra · 96 GB | Q8_0 | Runs great | 40 GB | 18 |
| M5 Ultra · 256 GB | Q8_0 | Runs great | 40 GB | 18 |
| M5 Ultra · 512 GB | Q8_0 | Runs great | 40 GB | 18 |
| M4 · 16 GB | IQ2_XXS | Runs well | 10 GB | 3.8 |
| M4 · 24 GB | Q3_K_S | Runs well | 16 GB | 3 |
| M4 · 32 GB | Q4_K_M | Runs well | 21 GB | 2.4 |
| M4 Pro · 24 GB | Q3_K_S | Runs well | 16 GB | 6.4 |
| M4 Pro · 48 GB | Q4_K_L | Runs great | 23 GB | 5.2 |
| M4 Pro · 64 GB | Q6_K_L | Runs great | 30 GB | 4.3 |
| M4 Max · 36 GB | Q4_K_L | Runs well | 23 GB | 7.5 |
| M4 Max · 48 GB | Q4_K_L | Runs great | 23 GB | 9.6 |
| M4 Max · 64 GB | Q6_K_L | Runs great | 30 GB | 8 |
| M4 Max · 96 GB | Q8_0 | Runs great | 40 GB | 5.9 |
| M4 Max · 128 GB | Q8_0 | Runs great | 40 GB | 7.7 |
| M3 · 8 GB | IQ2_XXS | Too big | 14 GB | |
| M3 · 16 GB | IQ2_XXS | Runs well | 10 GB | 3.3 |
| M3 · 24 GB | Q3_K_S | Runs well | 16 GB | 2.5 |
| M3 Pro · 18 GB | IQ2_S | Runs well | 12 GB | 4.5 |
| M3 Pro · 36 GB | Q4_K_L | Runs well | 23 GB | 3 |
| M3 Max · 36 GB | Q4_K_L | Runs well | 23 GB | 5.7 |
| M3 Max · 48 GB | Q4_K_L | Runs great | 23 GB | 7.4 |
| M3 Max · 64 GB | Q6_K_L | Runs great | 30 GB | 6.2 |
| M3 Max · 96 GB | Q8_0 | Runs great | 40 GB | 4.5 |
| M3 Max · 128 GB | Q8_0 | Runs great | 40 GB | 5.8 |
| M3 Ultra · 96 GB | Q8_0 | Runs great | 40 GB | 11 |
| M3 Ultra · 256 GB | Q8_0 | Runs great | 40 GB | 11 |
| M3 Ultra · 512 GB | Q8_0 | Runs great | 40 GB | 11 |
| M2 · 8 GB | IQ2_XXS | Too big | 14 GB | |
| M2 · 16 GB | IQ2_XXS | Runs well | 10 GB | 3.3 |
| M2 · 24 GB | Q3_K_S | Runs well | 16 GB | 2.5 |
| M2 Pro · 16 GB | IQ2_XXS | Runs well | 10 GB | 6.3 |
| M2 Pro · 32 GB | Q4_K_M | Runs well | 21 GB | 4 |
| M2 Max · 32 GB | Q4_K_M | Runs well | 21 GB | 7.6 |
| M2 Max · 64 GB | Q6_K_L | Runs great | 30 GB | 6.1 |
| M2 Max · 96 GB | Q8_0 | Runs great | 40 GB | 5.8 |
| M2 Ultra · 64 GB | Q6_K_L | Runs great | 30 GB | 11 |
| M2 Ultra · 128 GB | Q8_0 | Runs great | 40 GB | 11 |
| M2 Ultra · 192 GB | Q8_0 | Runs great | 40 GB | 11 |
| M1 · 8 GB | IQ2_XXS | Too big | 14 GB | |
| M1 · 16 GB | IQ2_XXS | Runs well | 10 GB | 2.2 |
| M1 Pro · 16 GB | IQ2_XXS | Runs well | 10 GB | 5.9 |
| M1 Pro · 32 GB | Q4_K_M | Runs well | 21 GB | 3.9 |
| M1 Max · 32 GB | Q4_K_M | Runs well | 21 GB | 7.1 |
| M1 Max · 64 GB | Q6_K_L | Runs great | 30 GB | 5.8 |
| M1 Ultra · 64 GB | Q6_K_L | Runs great | 30 GB | 10 |
| M1 Ultra · 128 GB | Q8_0 | Runs great | 40 GB | 10 |
Writes is words a second, estimated from the Mac's memory speed and the model's size; how, and how wrong.
Packaged by
6 repositories on Hugging Face carry this model for llama.cpp. Outmake reads them all as one model.
Information
- Maker
- Alibaba Qwen
- Released
- November 2024
- Parameters
- 33B
- Longest chat
- 32K tokens
- Reads pictures
- No
- Licence
- apache-2.0
- Downloads this month
- 101,270
- Likes
- 402
- Sizes
- 21, 9.0 to 27 GB
- Kind
- Community version
Badge for the model card
Shows on Qwen2.5 Coder 32B's Hugging Face card which Macs run it well, linked to this page. It updates every morning.
Files, downloads and likes read from Hugging Face on November 5, 2026 and refreshed every morning. Memory and speed are estimates until someone running Outmake measures this model on that Mac.