Mistral Medium 3.5 128B VRAM requirements

How much VRAM does Mistral Medium 3.5 128B need?

Mistral Medium 3.5 128B needs 82.7 GB at Q4_K_M, 143 GB at Q8_0 and 265 GB at FP16 with 8,192 tokens of context and one request. Each total below is the weights plus the FP16 KV cache and the runtime overhead (0.5 GB plus 10%), and each row opens the calculator with that setting. What the GGUF names mean.

Fine-tuning Mistral Medium 3.5 128B? Mistral Medium 3.5 128B VRAM for LoRA, QLoRA and full training.

Mistral Medium 3.5 128B GGUF files on Hugging Face

The 8 GGUF files llama.cpp picks for Mistral Medium 3.5 128B with -hf <repo>:<QUANT>, with the bytes the Hugging Face file list reported on 2026-09-29. Each total is that file plus the FP16 KV cache and 0.5 GB + 10%, one request. "vs estimate" compares the file with the bits-per-weight size the table above uses.

GB here is GiB (1024³ bytes), as everywhere on this site; Hugging Face shows the Q4_K_M file as 74.9 GB, in 10⁹ bytes.

Which Mistral Medium 3.5 128B GGUF fits a 16, 24, 32, 48 or 80 GB GPU or a Mac

From the file sizes above: the largest file that fits each machine with an 8K context and 0.5 GB left free, the longest context it then has room for, and the longest context for Q4_K_M (69.8 GB). One request, FP16 KV cache. The Q4_K_M file is 2.20 GB under its bits-per-weight estimate; the estimate tables and the llama-server command on this page count the estimate, so they show less context.

Run Mistral Medium 3.5 128B with llama.cpp

llama-server -hf unsloth/Mistral-Medium-3.5-128B-GGUF:Q4_K_M -c 41984

Q4_K_M/Mistral-Medium-3.5-128B-Q4_K_M-00001-of-00003.gguf, 74.9 GB, from unsloth/Mistral-Medium-3.5-128B-GGUF (checked 2026-09-29). -c 41984 is the longest Q4_K_M context on the Ryzen AI Max+ 395 (128 GB, 96 GB usable).

Mistral Medium 3.5 128B at 8K, 32K, 128K and 256K (full) tokens of context

All 88 layers use full attention. Each extra token of context adds 352 KB of FP16 cache per request. The last column is the smallest setup here, one card or a group, that holds it at Q4_K_M. How the KV cache works.

How fast Mistral Medium 3.5 128B writes

Tokens per second for one request with 8,192 tokens of context, estimated from memory bandwidth. A dash means it does not fit on one card. Try other settings in the speed calculator.

Longest context on one GPU

How many tokens of context Mistral Medium 3.5 128B fits on a single card with one request, an FP16 KV cache and 0.5 GB left free. "Full" means the model's whole context window fits.

Model details

Parameters

127.7B (127,704,210,176)

Layers

All 88 layers use full attention

Attention cache

8 KV heads × 128

Context length

262,144 tokens

Published weights

124 GB (FP8)

On Hugging Face

mistralai/Mistral-Medium-3.5-128B

Reported benchmarks

None published in the model card for GPQA, MMLU-Pro or LiveCodeBench.

Mistral Medium 3.5 128B next to similar-sized models

The three models here closest to it in parameter count, at Q4_K_M with 8,192 tokens of context.