Mistral Large 3 675B
By Mistral AI · France
Updated 2026-07-13
Overview
Mistral AI's flagship 675B MoE (41B active) with a 2.5B vision encoder, trained from scratch on 3,000 H200s and released under Apache 2.0. Currently #2 OSS non-reasoning model on LMArena.
When to pick this model
- Frontier-tier on-prem deployments needing permissive licensing
- Multimodal applications requiring top-tier text quality
- Sovereign or regulated environments that cannot ship data to closed APIs
- Multilingual production workloads across European languages
- Replacing GPT-4-class APIs in self-hosted stacks
VRAM requirements by quantization
| Quantization | VRAM required |
|---|---|
| Q4_K_M (recommended) | 405 GB |
| Q5_K_M | 485 GB |
| Q8_0 | 720 GB |
| FP16 (no quantization) | 1350 GB |
VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.
In practice, Mistral Large 3 675B is server-class even at Q4_K_M (405 GB). Stepping up to Q8_0 nearly doubles the footprint to 720 GB, and unquantized FP16 weights take 1350 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.
Without a GPU, Mistral Large 3 675B needs roughly 480 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 1 tokens/sec on entry-level GPUs, on the order of 5 tokens/sec on a mid-range card, and up to 15 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.
What hardware do you need
The table below matches Mistral Large 3 675B to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.
| GPU memory | Example cards | Best fit for Mistral Large 3 675B |
|---|---|---|
| 8 GB | RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB | Does not fit — needs 405 GB at Q4_K_M |
| 12 GB | RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop | Does not fit — needs 405 GB at Q4_K_M |
| 16 GB | RTX 5080, RTX 4080 Super, Radeon RX 9070 XT | Does not fit — needs 405 GB at Q4_K_M |
| 24 GB | RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop | Does not fit — needs 405 GB at Q4_K_M |
| 32 GB | RTX 5090 | Does not fit — needs 405 GB at Q4_K_M |
Which GPU should you buy to run Mistral Large 3 675B?
To run Mistral Large 3 675B locally at Q4, you need ~405 GB of VRAM. The best value for this is a Apple Mac Studio (64+ GB unified memory).
As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.
Strengths
- Top-tier open weights — #2 OSS non-reasoning on LMArena
- Apache 2.0 — fully unrestricted commercial use
- Native multimodal with 2.5B vision encoder
- 256k context window
- Strong multilingual coverage out of the box
Limitations
- 405GB at Q4 — needs an H200 or B200 server class deployment
- Active expert count (41B) still demands substantial inference compute
- Overkill for most single-GPU or developer-laptop use cases
Typical workloads
In our catalog grid, Mistral Large 3 675B is filed under Open EU Frontier, Vision, Multilingual — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: vision-language work — screenshots, charts, scanned documents; multilingual workloads; French-language output where quality matters.
The 250k-token context window is large enough to hold entire codebases' worth of files or long reports in a single prompt, which is what makes local RAG and document analysis practical. The Apache 2.0 license is permissive, so shipping it inside a commercial product raises no special legal questions.
Architecture & training
Architecture: Granular MoE 675B/41B active + 2.5B vision encoder · 256k ctx
Training: From scratch on 3000 H200.
The most capable open-weight non-reasoning model shipping today — if you have the H200s, this replaces closed frontier APIs.
Quick start
# HuggingFace : mistralai/Mistral-Large-3-675B-Instruct-2512Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.
Similar models worth comparing
Frequently asked questions
How much VRAM does Mistral Large 3 675B need?
At the recommended Q4_K_M quantization, Mistral Large 3 675B needs about 405 GB of VRAM. Q8_0 takes 720 GB, and unquantized FP16 weights take 1350 GB.
Can Mistral Large 3 675B run without a GPU?
Yes — with roughly 480 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.
What context window does Mistral Large 3 675B support?
Mistral Large 3 675B supports a 250k-token context window (256,000 tokens).
Can I use Mistral Large 3 675B commercially?
Yes. Mistral Large 3 675B is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.
How fast is Mistral Large 3 675B on consumer hardware?
Our compatibility engine estimates on the order of 5 tokens/sec on a mid-range GPU and up to 15 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.
Which quantization of Mistral Large 3 675B should I download first?
Start with Q4_K_M (405 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It does not fit a single 24 GB consumer card — plan for multi-GPU or server hardware.