Hugging FaceA 299B, 17B active chat model from Tencent.
- Rating
- 4.6
- Downloads
- 147K
- Likes
- 197
- Parameters
- 299B, 17B active
- Context
- 256K
- Licence
- apache-2.0
Too big for this Mac
Its smallest size needs 114 GB of memory, more than M4 · 16 GB can give it.
Ratings
4.6 out of 5, across people's blind votes on its answers and the tests below.
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.
GPQA Diamond: questions PhD scientists write for each other. Humanity's Last Exam: questions written to be beyond today's models. SWE-bench Verified: real GitHub issues fixed so the project's tests pass. Scores are the ones filed on the model's Hugging Face page or the board's own table, read 2026-11-05; some come from the maker.
Hy3 runs well from a Mac with an M2 Max and 96 GB of memory. 15 of 59 Mac configurations run it well.
Compare with
The nearest models in size.
- MiMo V2 FlashXiaomi · 21K downloads · Arena 1386Too big for this Mac · 78 GB
- MiMo V2.6 Flash RLXiaomi · 47K downloadsToo big for this Mac · 108 GB
- MiMo V2.5Xiaomi · 3.6K downloads · Arena 1434Too big for this Mac · 93 GB
- Llama 3 1 Nemotron Ultra 253B V1NVIDIA · 2.2K downloads · Arena 1348Too big for this Mac · 54 GB
Sizes
From bartowski/Hy3-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.
| Size | Download | Memory at 8K | Get |
|---|---|---|---|
| IQ1_S | 64 GB | 66 GBestimated | Open |
| IQ1_M | 71 GB | 73 GBestimated | Open |
| IQ2_XXS | 82 GB | 84 GBestimated | Open |
| IQ2_XS | 91 GB | 93 GBestimated | Open |
| IQ2_S | 93 GB | 94 GBestimated | Open |
| IQ2_M | 102 GB | 104 GBestimated | Open |
| Q2_K | 107 GB | 108 GBestimated | Open |
| Q2_K_L | 107 GB | 109 GBestimated | Open |
| IQ3_XXS | 126 GB | 128 GBestimated | Open |
| Q3_K_S | 131 GB | 133 GBestimated | Open |
| IQ3_XS | 137 GB | 139 GBestimated | Open |
| Q3_K_M | 138 GB | 139 GBestimated | Open |
| Q3_K_L | 143 GB | 145 GBestimated | Open |
| Q3_K_XL | 144 GB | 145 GBestimated | Open |
| IQ3_M | 144 GB | 145 GBestimated | Open |
| IQ4_XS | 161 GB | 162 GBestimated | Open |
| IQ4_NL | 170 GB | 171 GBestimated | Open |
| Q4_0 | 170 GB | 172 GBestimated | Open |
| Q4_K_S | 175 GB | 177 GBestimated | Open |
| Q4_K_M | 182 GB | 184 GBestimated | Open |
| Q4_K_L | 183 GB | 184 GBestimated | Open |
| Q4_1 | 188 GB | 189 GBestimated | Open |
| Q5_K_S | 206 GB | 208 GBestimated | Open |
| Q5_K_M | 213 GB | 215 GBestimated | Open |
| Q6_K | 257 GB | 259 GBestimated | Open |
| Q8_0 | 318 GB | 319 GBestimated | Open |
Run It Anywhere
Hy3 is an open model, so any app that runs GGUF files can run it. These take the IQ1_S 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/Hy3-GGUF:IQ1_S - llama.cppMac, Windows, Linux
llama-server -hf bartowski/Hy3-GGUF:IQ1_S - Docker Model RunnerMac, Windows, Linux
docker model run hf.co/bartowski/Hy3-GGUF:IQ1_S
On a PC Graphics Card
| Video memory | Largest size that fits | Download | Memory at 8K |
|---|---|---|---|
| 6 GB | None worth running | needs 128 GB | |
| 8 GB | None worth running | needs 128 GB | |
| 10 GB | None worth running | needs 128 GB | |
| 11 GB | None worth running | needs 128 GB | |
| 12 GB | None worth running | needs 128 GB | |
| 16 GB | None worth running | needs 128 GB | |
| 20 GB | None worth running | needs 128 GB | |
| 24 GB | None worth running | needs 128 GB | |
| 32 GB | None worth running | needs 128 GB | |
| 48 GB | None worth running | needs 128 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 | IQ1_S | Too big | 114 GB | |
| M6 · 24 GB | IQ1_S | Too big | 114 GB | |
| M6 · 32 GB | IQ1_S | Too big | 114 GB | |
| M5 · 16 GB | IQ1_S | Too big | 114 GB | |
| M5 · 24 GB | IQ1_S | Too big | 114 GB | |
| M5 · 32 GB | IQ1_S | Too big | 114 GB | |
| M5 Pro · 24 GB | IQ1_S | Too big | 114 GB | |
| M5 Pro · 48 GB | IQ1_S | Too big | 114 GB | |
| M5 Pro · 64 GB | IQ1_S | Too big | 114 GB | |
| M5 Max · 36 GB | IQ1_S | Too big | 114 GB | |
| M5 Max · 48 GB | IQ1_S | Too big | 114 GB | |
| M5 Max · 64 GB | IQ1_S | Too big | 114 GB | |
| M5 Max · 128 GB | IQ1_S | Runs great | 67 GB | 16 |
| M5 Ultra · 96 GB | IQ1_S | Runs well | 67 GB | 28 |
| M5 Ultra · 256 GB | Q4_0 | Runs great | 174 GB | 24 |
| M5 Ultra · 512 GB | Q8_0 | Runs great | 343 GB | 18 |
| M4 · 16 GB | IQ1_S | Too big | 114 GB | |
| M4 · 24 GB | IQ1_S | Too big | 114 GB | |
| M4 · 32 GB | IQ1_S | Too big | 114 GB | |
| M4 Pro · 24 GB | IQ1_S | Too big | 114 GB | |
| M4 Pro · 48 GB | IQ1_S | Too big | 114 GB | |
| M4 Pro · 64 GB | IQ1_S | Too big | 114 GB | |
| M4 Max · 36 GB | IQ1_S | Too big | 114 GB | |
| M4 Max · 48 GB | IQ1_S | Too big | 114 GB | |
| M4 Max · 64 GB | IQ1_S | Too big | 114 GB | |
| M4 Max · 96 GB | IQ1_S | Runs well | 67 GB | 9 |
| M4 Max · 128 GB | IQ1_S | Runs great | 67 GB | 11 |
| M3 · 8 GB | IQ1_S | Too big | 114 GB | |
| M3 · 16 GB | IQ1_S | Too big | 114 GB | |
| M3 · 24 GB | IQ1_S | Too big | 114 GB | |
| M3 Pro · 18 GB | IQ1_S | Too big | 114 GB | |
| M3 Pro · 36 GB | IQ1_S | Too big | 114 GB | |
| M3 Max · 36 GB | IQ1_S | Too big | 114 GB | |
| M3 Max · 48 GB | IQ1_S | Too big | 114 GB | |
| M3 Max · 64 GB | IQ1_S | Too big | 114 GB | |
| M3 Max · 96 GB | IQ1_S | Runs well | 67 GB | 7.2 |
| M3 Max · 128 GB | IQ1_S | Runs great | 67 GB | 9.1 |
| M3 Ultra · 96 GB | IQ1_S | Runs well | 67 GB | 15 |
| M3 Ultra · 256 GB | Q4_0 | Runs great | 174 GB | 13 |
| M3 Ultra · 512 GB | Q8_0 | Runs great | 343 GB | 10 |
| M2 · 8 GB | IQ1_S | Too big | 114 GB | |
| M2 · 16 GB | IQ1_S | Too big | 114 GB | |
| M2 · 24 GB | IQ1_S | Too big | 114 GB | |
| M2 Pro · 16 GB | IQ1_S | Too big | 114 GB | |
| M2 Pro · 32 GB | IQ1_S | Too big | 114 GB | |
| M2 Max · 32 GB | IQ1_S | Too big | 114 GB | |
| M2 Max · 64 GB | IQ1_S | Too big | 114 GB | |
| M2 Max · 96 GB | IQ1_S | Runs well | 67 GB | 9.1 |
| M2 Ultra · 64 GB | IQ1_S | Too big | 114 GB | |
| M2 Ultra · 128 GB | IQ1_S | Runs great | 67 GB | 15 |
| M2 Ultra · 192 GB | IQ3_XXS | Runs great | 129 GB | 13 |
| M1 · 8 GB | IQ1_S | Too big | 114 GB | |
| M1 · 16 GB | IQ1_S | Too big | 114 GB | |
| M1 Pro · 16 GB | IQ1_S | Too big | 114 GB | |
| M1 Pro · 32 GB | IQ1_S | Too big | 114 GB | |
| M1 Max · 32 GB | IQ1_S | Too big | 114 GB | |
| M1 Max · 64 GB | IQ1_S | Too big | 114 GB | |
| M1 Ultra · 64 GB | IQ1_S | Too big | 114 GB | |
| M1 Ultra · 128 GB | IQ1_S | Runs great | 67 GB | 12 |
Writes is words a second, estimated from the Mac's memory speed and the model's size; how, and how wrong.
Packaged by
2 repositories on Hugging Face carry this model for llama.cpp. Outmake reads them all as one model.
Information
- Maker
- Tencent
- Released
- July 2026
- Parameters
- 299B, 17B active
- Longest chat
- 256K tokens
- Reads pictures
- No
- Licence
- apache-2.0
- Downloads this month
- 146,727
- Likes
- 197
- Sizes
- 26, 64 to 318 GB
- Kind
- Maker's release
- Source
- tencent/Hy3
Badge for the model card
Shows on Hy3'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.