// Open-model VRAM table · family
Mistral / Magistral / Devstral: local deployment & VRAM
35 open-weight models from Mistral AI (61 official repos incl. quantised builds). At 4-bit they range from 4.2 GB (Ministral-3-3B-Instruct-2512-BF16) to 451 GB (Mistral-Large-3-675B-Instruct-2512).
official orgs: mistralailatest 2026-10-08memory = estimate
01Text LLM (30)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| LIDstral-Arabic | — | — | — | — | — | — | Commercial OK | 2026-10-08 |
| Mistral-Medium-3.5-128B | 127.7B | 256K | 284 GB | 88.6 GB | H200 141GB | Mac 256GB | Custom licence – read it | 2026-03-31 |
| Mistral-Small-4-119B-2603 | 119B / 6.5B | 1M | 263 GB | 80.0 GB | RTX PRO 6000 96GB | Mac 128GB | Commercial OK | 2026-01-23 |
| Devstral-2-123B-Instruct-2512 | 125B | 256K | 279 GB | 86.8 GB | H200 141GB | Mac 256GB | Custom licence – read it | 2025-11-28 |
| Mistral-Large-3-675B-Instruct-2512 | 675B / 41B | 288K | 1.49 TB | 451 GB | 8 × H100 (640GB) | — | Commercial OK | 2025-11-28 |
| Ministral-3-3B-Instruct-2512-BF16 | 4.25B | 256K | 10.7 GB | 4.2 GB | 8GB GPU | Mac 16GB | Commercial OK | 2025-10-31 |
| Ministral-3-8B-Instruct-2512-BF16 | 8.92B | 256K | 21.3 GB | 7.6 GB | 12GB GPU | Mac 16GB | Commercial OK | 2025-10-31 |
| Ministral-3-14B-Instruct-2512-BF16 | 13.9B | 256K | 32.5 GB | 11.1 GB | 16GB GPU | Mac 24GB | Commercial OK | 2025-10-31 |
| Ministral-3-3B-Reasoning-2512 | 4.25B | 256K | 10.7 GB | 4.2 GB | 8GB GPU | Mac 16GB | Commercial OK | 2025-10-31 |
| Ministral-3-8B-Reasoning-2512 | 8.92B | 256K | 21.3 GB | 7.6 GB | 12GB GPU | Mac 16GB | Commercial OK | 2025-10-31 |
| Ministral-3-14B-Reasoning-2512 | 13.9B | 256K | 32.5 GB | 11.1 GB | 16GB GPU | Mac 24GB | Commercial OK | 2025-10-31 |
| Magistral-Small-2509 | 24B | 128K | 54.7 GB | 17.9 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-09-12 |
| Magistral-Small-2507 | 23.6B | 40K | 53.7 GB | 17.6 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-07-18 |
| Devstral-Small-2507 | 23.6B | 128K | 53.7 GB | 17.6 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-07-04 |
| Voxtral-Mini-3B-2507 | 4.68B | 128K | 11.8 GB | 4.6 GB | 8GB GPU | Mac 16GB | Commercial OK | 2025-07-01 |
| Mistral-Small-3.2-24B-Instruct-2506 | 24B | 128K | 54.7 GB | 17.9 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-06-19 |
| Magistral-Small-2506 | 23.6B | 40K | 53.7 GB | 17.6 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-06-04 |
| Devstral-Small-2505 | 23.6B | 128K | 53.7 GB | 17.6 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-05-12 |
| Mistral-Small-3.1-24B-Instruct-2503 | 24B | 128K | 54.7 GB | 17.9 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-03-11 |
| Mistral-Small-24B-Instruct-2501 | 23.6B | 32K | 53.7 GB | 17.6 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-01-28 |
| Mistral-Large-Instruct-2411 | 122.6B | 128K | 273 GB | 85.2 GB | RTX PRO 6000 96GB | Mac 128GB | Non-commercial | 2024-11-14 |
| Ministral-8B-Instruct-2410 | 8.02B | 32K | 19.4 GB | 7.1 GB | 8GB GPU | Mac 16GB | Non-commercial | 2024-10-15 |
| Mistral-Small-Instruct-2409 | 22.2B | 32K | 51.3 GB | 17.2 GB | RTX 4090 24GB | Mac 36GB | Non-commercial | 2024-09-17 |
| Pixtral-12B-2409 | 12B | 128K | 28.2 GB | 9.8 GB | 12GB GPU | Mac 24GB | Commercial OK | 2024-09-11 |
| Mistral-Large-Instruct-2407 | 122.6B | — | 270 GB | 82.3 GB | RTX PRO 6000 96GB | Mac 128GB | Non-commercial | 2024-07-24 |
| Mistral-Nemo-Instruct-2407 | 12.2B | 128K | 28.8 GB | 10.0 GB | 12GB GPU | Mac 24GB | Commercial OK | 2024-07-17 |
| Codestral-22B-v0.1 | 22.2B | 32K | 51.3 GB | 17.2 GB | RTX 4090 24GB | Mac 36GB | Non-commercial | 2024-05-29 |
| Mistral-7B-Instruct-v0.3 | 7.25B | 32K | 17.5 GB | 6.4 GB | 8GB GPU | Mac 16GB | Commercial OK | 2024-05-22 |
| Mixtral-8x22B-Instruct-v0.1 | 140.6B | 64K | 312 GB | 96.2 GB | H200 141GB | Mac 256GB | Commercial OK | 2024-04-16 |
| Mixtral-8x7B-Instruct-v0.1 | 46.7B | 32K | 104 GB | 32.7 GB | 2 × RTX 5090 (64GB) | Mac 64GB | Commercial OK | 2023-12-10 |
02Omni / audio LLM (1)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| Voxtral-Small-24B-2507 | 24.3B | 128K | 55.2 GB | 18.0 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-07-01 |
03Speech-to-text (2)
| Model | Params | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|
| Voxtral-Mini-4B-Realtime-Arabic | 4.43B | 10.2 GB | 3.5 GB | 8GB GPU | Mac 16GB | Licence not stated | 2026-10-08 |
| Voxtral-Mini-4B-Realtime-2602 | 4.43B | 10.2 GB | 3.5 GB | 8GB GPU | Mac 16GB | Commercial OK | 2026-01-21 |
04Text-to-speech / audio (1)
| Model | Params | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|
| Voxtral-4B-TTS-2603 ↗ | 4B | 9.3 GB | 3.2 GB | 8GB GPU | Mac 16GB | Non-commercial | 2025-11-17 |
05Safety / guard (1)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| Shieldstral-1.0-3B | 3.85B | 256K | 9.8 GB | 3.9 GB | 8GB GPU | Mac 16GB | Commercial OK | 2026-07-16 |
06Other families
Qwen · IBM Granite · NVIDIA Nemotron · Gemma · GLM / CogVideoX (Z.ai) · DeepSeek · Ai2 OLMo / Molmo · Ant Ling / Ring / Ming · Liquid LFM · MiniCPM / VoxCPM · Tencent Hunyuan / Hy · Phi · Llama · Meituan LongCat · Jina embeddings & rerankers · Wan · Kimi · Cohere Command / Aya · Xiaomi MiMo · FLUX · Microsoft VibeVoice / Fara / Harrier · InternLM / Intern-S · StepFun Step · MiniMax · Baidu ERNIE · NVIDIA Parakeet / Canary speech · Stable Diffusion / Stable Audio · Yi (01.AI) · ByteDance Seed · SmolLM / SmolVLM · BGE embeddings & rerankers · LTX Video · FunAudioLLM speech · gpt-oss · poolside Laguna · Whisper · Baichuan · Thinking Machines Inkling · xAI Grok · rednote dots
Totals include 10% + 0.5GB overhead and one 8K-token KV cache for text models. Estimates only. Method →