Yes, just — 10.7 GB spare

Mistral Small 4 119B-A6B at Q4_K_M fits your A100 80 GB entirely on the GPU at 8K context, at an estimated 130 tokens per second. There is room for its full 256K window.

Fully on GPU 8K context Q4_K_M · 66.9 GB Apache 2.0 Released 17 Mar 2026 Vision Not in the Ollama library

Instruct, reasoning, vision and code in one 119B MoE. A latent KV cache (320 wide) keeps context cheap; the weights still want 64 GB+.

What hardware do I need for Mistral Small 4 119B-A6B? →

The VRAM budget

weights 66.9 GB

Weights 66.9 GB KV cache @ 8K 0.18 GB Runtime overhead 0.6 GB Free 10.7 GB of 78.4 GB

Quantisation ladder

Quant	Weights	Total @ 8K	Max context	Tok/s	Quality	Fit
Q8_0	117.8 GB	118.5 GB	—	~6.0	−0.1% ppl	40.1 GB over
Q6_K	90.9 GB	91.7 GB	—	~16	−0.4% ppl	13.3 GB over
Q5_K_M	78.5 GB	79.3 GB	—	~80	−0.8% ppl	0.9 GB over
Q4_K_M	66.9 GB	67.7 GB	256K	130	−1.9% ppl	Recommended
Q3_K_M	54.2 GB	54.9 GB	256K	160	−5.4% ppl	Long context
Q2_K	46.4 GB	47.2 GB	256K	187	−15% ppl	Long context

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model uses multi-head latent attention, so its cache is a compressed latent rather than full K and V.

How to run it

llama.cpp LM Studio vLLM

terminal

$ llama-server \
    -hf mistralai/Mistral-Small-4-119B-2603:Q4_K_M \
    -c 8192 -ngl 99

The engine underneath most of the others. Every knob is exposed. More on llama.cpp.

01Download is 66.9 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.

02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.

03Only 10.7 GB is spare, so a long system prompt can still push it over. Its real ceiling here is 256K context.

See all models for this rig Compare with another model