Hugging FaceA 7.3B chat model from Mistral AI.
- Downloads
- 105K
- Likes
- 262
- Parameters
- 7.3B
- Context
- 32K
- Licence
- apache-2.0
Runs great on M4 · 16 GB
The Q5_K size from legraphista is the one to get: a 5.1 GB download that uses 7.7 GB of memory, and writes about 7.2 words a second.
Mistral 7B V0.3 runs well from a Mac with an M1 and 8 GB of memory. 59 of 59 Mac configurations run it well.
Compare with
The nearest models in size.
- Mathstral 7B V0.1Mistral AI · 44K downloadsRuns great · 6.0 GB · 6.6 words/s
- Mistral 7B V0.1Mistral AI · 60K downloadsRuns great · 5.9 GB · 6.7 words/s
- Mistral 7B V0.2Mistral AI · 46K downloads · Arena 1149Runs great · 5.9 GB · 6.7 words/s
- Gemma 4 E4B Qat Mobileunsloth · 52K downloadsRuns great · 3.2 GB · 14 words/s
Sizes
From legraphista/Mistral-7B-v0.3-IMat-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.
| Size | Download | Memory at 8K | Get |
|---|---|---|---|
| IQ1_S | 1.6 GB | 2.3 GBestimated | Open |
| IQ1_M | 1.8 GB | 2.5 GBestimated | Open |
| IQ2_XXS | 2.0 GB | 2.7 GBestimated | Open |
| IQ2_XS | 2.2 GB | 2.9 GBestimated | Open |
| IQ2_S | 2.3 GB | 3.0 GBestimated | Open |
| IQ2_M | 2.5 GB | 3.2 GBestimated | Open |
| Q2_K_S | 2.5 GB | 3.3 GBestimated | Open |
| Q2_K | 2.7 GB | 3.5 GBestimated | Open |
| IQ3_XXS | 2.8 GB | 3.6 GBestimated | Open |
| IQ3_XS | 3.0 GB | 3.8 GBestimated | Open |
| Q3_K_S | 3.2 GB | 3.9 GBestimated | Open |
| IQ3_S | 3.2 GB | 3.9 GBestimated | Open |
| IQ3_M | 3.3 GB | 4.0 GBestimated | Open |
| Q3_K | 3.5 GB | 4.3 GBestimated | Open |
| Q3_K_L | 3.8 GB | 4.6 GBestimated | Open |
| IQ4_XS | 3.9 GB | 4.6 GBestimated | Open |
| IQ4_NL | 4.1 GB | 4.9 GBestimated | Open |
| Q4_K_S | 4.1 GB | 4.9 GBestimated | Open |
| Q4_K | 4.4 GB | 5.1 GBestimated | Open |
| Q5_K_S | 5.0 GB | 5.7 GBestimated | Open |
| Q5_K | 5.1 GB | 5.9 GBestimated | Open |
Run It Anywhere
Mistral 7B V0.3 is an open model, so any app that runs GGUF files can run it. These take the Q5_K size from legraphista straight from Hugging Face.
- LM StudioMac, Windows, Linux
Open - JanMac, Windows, Linux
Open - MstyMac, Windows, Linux
Open - OllamaMac, Windows, Linux
ollama run hf.co/legraphista/Mistral-7B-v0.3-IMat-GGUF:Q5_K - llama.cppMac, Windows, Linux
llama-server -hf legraphista/Mistral-7B-v0.3-IMat-GGUF:Q5_K - Docker Model RunnerMac, Windows, Linux
docker model run hf.co/legraphista/Mistral-7B-v0.3-IMat-GGUF:Q5_K
On a PC Graphics Card
| Video memory | Largest size that fits | Download | Memory at 8K |
|---|---|---|---|
| 6 GB | Q5_K | 5.1 GB | 5.9 GB |
| 8 GB | Q5_K | 5.1 GB | 5.9 GB |
| 10 GB | Q5_K | 5.1 GB | 5.9 GB |
| 11 GB | Q5_K | 5.1 GB | 5.9 GB |
| 12 GB | Q5_K | 5.1 GB | 5.9 GB |
| 16 GB | Q5_K | 5.1 GB | 5.9 GB |
| 20 GB | Q5_K | 5.1 GB | 5.9 GB |
| 24 GB | Q5_K | 5.1 GB | 5.9 GB |
| 32 GB | Q5_K | 5.1 GB | 5.9 GB |
| 48 GB | Q5_K | 5.1 GB | 5.9 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 | Q5_K | Runs great | 7.7 GB | 10 |
| M6 · 24 GB | Q5_K | Runs great | 7.7 GB | 10 |
| M6 · 32 GB | Q5_K | Runs great | 7.7 GB | 10 |
| M5 · 16 GB | Q5_K | Runs great | 7.7 GB | 9.3 |
| M5 · 24 GB | Q5_K | Runs great | 7.7 GB | 9.3 |
| M5 · 32 GB | Q5_K | Runs great | 7.7 GB | 9.3 |
| M5 Pro · 24 GB | Q5_K | Runs great | 7.7 GB | 18 |
| M5 Pro · 48 GB | Q5_K | Runs great | 7.7 GB | 18 |
| M5 Pro · 64 GB | Q5_K | Runs great | 7.7 GB | 18 |
| M5 Max · 36 GB | Q5_K | Runs great | 7.7 GB | 27 |
| M5 Max · 48 GB | Q5_K | Runs great | 7.7 GB | 35 |
| M5 Max · 64 GB | Q5_K | Runs great | 7.7 GB | 35 |
| M5 Max · 128 GB | Q5_K | Runs great | 7.7 GB | 35 |
| M5 Ultra · 96 GB | Q5_K | Runs great | 7.7 GB | 66 |
| M5 Ultra · 256 GB | Q5_K | Runs great | 7.7 GB | 66 |
| M5 Ultra · 512 GB | Q5_K | Runs great | 7.7 GB | 66 |
| M4 · 16 GB | Q5_K | Runs great | 7.7 GB | 7.2 |
| M4 · 24 GB | Q5_K | Runs great | 7.7 GB | 7.2 |
| M4 · 32 GB | Q5_K | Runs great | 7.7 GB | 7.2 |
| M4 Pro · 24 GB | Q5_K | Runs great | 7.7 GB | 15 |
| M4 Pro · 48 GB | Q5_K | Runs great | 7.7 GB | 15 |
| M4 Pro · 64 GB | Q5_K | Runs great | 7.7 GB | 15 |
| M4 Max · 36 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M4 Max · 48 GB | Q5_K | Runs great | 7.7 GB | 27 |
| M4 Max · 64 GB | Q5_K | Runs great | 7.7 GB | 27 |
| M4 Max · 96 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M4 Max · 128 GB | Q5_K | Runs great | 7.7 GB | 27 |
| M3 · 8 GB | Q4_K | Runs well | 5.1 GB | 6.6 |
| M3 · 16 GB | Q5_K | Runs great | 7.7 GB | 6.1 |
| M3 · 24 GB | Q5_K | Runs great | 7.7 GB | 6.1 |
| M3 Pro · 18 GB | Q5_K | Runs great | 7.7 GB | 9 |
| M3 Pro · 36 GB | Q5_K | Runs great | 7.7 GB | 9 |
| M3 Max · 36 GB | Q5_K | Runs great | 7.7 GB | 17 |
| M3 Max · 48 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M3 Max · 64 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M3 Max · 96 GB | Q5_K | Runs great | 7.7 GB | 17 |
| M3 Max · 128 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M3 Ultra · 96 GB | Q5_K | Runs great | 7.7 GB | 38 |
| M3 Ultra · 256 GB | Q5_K | Runs great | 7.7 GB | 38 |
| M3 Ultra · 512 GB | Q5_K | Runs great | 7.7 GB | 38 |
| M2 · 8 GB | Q4_K | Runs well | 5.1 GB | 6.6 |
| M2 · 16 GB | Q5_K | Runs great | 7.7 GB | 6.1 |
| M2 · 24 GB | Q5_K | Runs great | 7.7 GB | 6.1 |
| M2 Pro · 16 GB | Q5_K | Runs great | 7.7 GB | 12 |
| M2 Pro · 32 GB | Q5_K | Runs great | 7.7 GB | 12 |
| M2 Max · 32 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M2 Max · 64 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M2 Max · 96 GB | Q5_K | Runs great | 7.7 GB | 22 |
| M2 Ultra · 64 GB | Q5_K | Runs great | 7.7 GB | 38 |
| M2 Ultra · 128 GB | Q5_K | Runs great | 7.7 GB | 38 |
| M2 Ultra · 192 GB | Q5_K | Runs great | 7.7 GB | 38 |
| M1 · 8 GB | Q4_K | Runs well | 5.1 GB | 4.5 |
| M1 · 16 GB | Q5_K | Runs great | 7.7 GB | 4.1 |
| M1 Pro · 16 GB | Q5_K | Runs great | 7.7 GB | 11 |
| M1 Pro · 32 GB | Q5_K | Runs great | 7.7 GB | 11 |
| M1 Max · 32 GB | Q5_K | Runs great | 7.7 GB | 20 |
| M1 Max · 64 GB | Q5_K | Runs great | 7.7 GB | 20 |
| M1 Ultra · 64 GB | Q5_K | Runs great | 7.7 GB | 33 |
| M1 Ultra · 128 GB | Q5_K | Runs great | 7.7 GB | 33 |
Writes is words a second, estimated from the Mac's memory speed and the model's size; how, and how wrong.
Packaged by
5 repositories on Hugging Face carry this model for llama.cpp. Outmake reads them all as one model.
Information
- Maker
- Mistral AI
- Released
- May 2024
- Parameters
- 7.3B
- Longest chat
- 32K tokens
- Reads pictures
- No
- Licence
- apache-2.0
- Downloads this month
- 105,472
- Likes
- 262
- Sizes
- 21, 1.6 to 5.1 GB
- Kind
- Maker's release
Badge for the model card
Shows on Mistral 7B V0.3'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.