Mistral Large 3 675B Instruct

675B MoE

Mistral AI

Mistral 3 flagship MoE (41B active / 675B total) with vision encoder. FP8 on 8×H200; GGUF quant for research clusters only.

⬇ 1.1K HF downloads♥ 241 likesmistralai/Mistral-Large-3-675B-Instruct-2512· stats from 8/8/2026
Pro GPU

262K

Max Context

2

Quant Variants

GGUF Q4_K_M

Best Quality

97.9%

Accuracy Retained

Calculate VRAMHugging FaceCompare

Quantization Variants

Per-quant VRAM, quality loss, and inference speed on RTX 4090

Measured = site benchmarks · Estimated = formula · Community = public reports

FormatLevelBPWVRAMPPL LossSpeedSourceActions
GGUFQ4_K_M4.85388.0 GB2.1%4 tok/sEstimated
CalcHF
GGUFQ3_K_M3.87312.0 GB4.5%5 tok/sEstimated
CalcHF

Similar models

Compare with Mixtral 8x7B
47B MoE

Mixtral 8x7B Instruct

Mistral AI

Pro GPU
25.2 GBmin VRAM·97.2%accuracy

Classic MoE model. ~13B active params per token; needs 32GB+ VRAM for Q4.

24B

Mistral Small 24B Instruct

Mistral AI

Consumer GPUPro GPU
13.5 GBmin VRAM·97.8%accuracy

Mistral's efficient 24B. Strong multilingual; fits on 24GB with Q4.

22B

Codestral 22B

Mistral AI

Consumer GPUPro GPU
13.2 GBmin VRAM·97.0%accuracy

Mistral's dedicated code model. 80+ language support, Fill-in-the-Middle capable.

12B

Mistral Nemo 12B Instruct

Mistral AI

Consumer GPUMac / Apple Silicon
7.8 GBmin VRAM·99.1%accuracy

Mistral + NVIDIA collaboration. 128K context, excellent multilingual support.