BestLLMfor EN Your hardware. Your LLM. Your call.
APIOpen data Find my LLM
Model fiche

Mistral Small 4

By Mistral AI · France

Updated 2026-07-13

chat general code vision reasoning multilingual fr moe
Parameters
119B
License
Apache 2.0
Context
250k
VRAM (Q4)
72 GB
Released
March 2026

Overview

Mistral AI's 2026 flagship MoE with 119B total and 6.5B active parameters, unifying chat, reasoning, vision, and code in a single Apache 2.0 model.

When to pick this model

  • Consolidating multiple Mistral deployments into one model
  • Vision plus reasoning workloads on a prosumer rig
  • Long-context analysis up to 256K tokens
  • European-data-sovereignty deployments
  • Apache-licensed commercial products

VRAM requirements by quantization

VRAM REQUIRED (GB)24324880128Q4_K_M72 GBQ5_K_M86 GBQ8_0128 GBFP16238 GB
QuantizationVRAM required
Q4_K_M (recommended)72 GB
Q5_K_M86 GB
Q8_0128 GB
FP16 (no quantization)238 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 Small 4 is server-class even at Q4_K_M (72 GB). Stepping up to Q8_0 nearly doubles the footprint to 128 GB, and unquantized FP16 weights take 238 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.

Without a GPU, Mistral Small 4 needs roughly 96 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 3 tokens/sec on entry-level GPUs, on the order of 12 tokens/sec on a mid-range card, and up to 30 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.

What hardware do you need

The table below matches Mistral Small 4 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 memoryExample cardsBest fit for Mistral Small 4
8 GBRTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GBDoes not fit — needs 72 GB at Q4_K_M
12 GBRTX 5070, RTX 5070 Ti Laptop, RTX 4080 LaptopDoes not fit — needs 72 GB at Q4_K_M
16 GBRTX 5080, RTX 4080 Super, Radeon RX 9070 XTDoes not fit — needs 72 GB at Q4_K_M
24 GBRTX 4090, Radeon RX 7900 XTX, RTX 5090 LaptopDoes not fit — needs 72 GB at Q4_K_M
32 GBRTX 5090Does not fit — needs 72 GB at Q4_K_M

Which GPU should you buy to run Mistral Small 4?

To run Mistral Small 4 locally at Q4, you need ~72 GB of VRAM. The best value for this is a Apple Mac Studio (64+ GB unified memory).

Check Apple Mac Studio price on Amazon →

As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.

Strengths

  • Unifies chat, reasoning, vision, and code in one model
  • Only 6.5B active parameters for fast inference
  • 256K context window
  • Apache 2.0 license
  • European lab with strong French and EU-language support

Limitations

  • 72GB+ in Q4 requires a prosumer multi-GPU setup
  • Breaks continuity with the Small 3.x line
  • Newer release means thinner ecosystem

Typical workloads

In our catalog grid, Mistral Small 4 is filed under Agents, Vision, Code, Reasoning — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: code generation and review (pair it with an editor integration like Continue.dev or Cline); multi-step reasoning and math-flavoured tasks; 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: MoE 119B/6.5B active · 256k ctx · unifies instruct+reasoning+vision+code

Training: Replaces Small 3.x and Pixtral in a single model.

Verdict

Mistral's most ambitious open release yet, ideal if you want one model covering four product lines.

Quick start

# HuggingFace : mistralai/Mistral-Small-4 (pas encore de tag Ollama officiel)

Or 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 Small 4 need?

At the recommended Q4_K_M quantization, Mistral Small 4 needs about 72 GB of VRAM. Q8_0 takes 128 GB, and unquantized FP16 weights take 238 GB.

Can Mistral Small 4 run without a GPU?

Yes — with roughly 96 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 Small 4 support?

Mistral Small 4 supports a 250k-token context window (256,000 tokens).

Can I use Mistral Small 4 commercially?

Yes. Mistral Small 4 is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.

How fast is Mistral Small 4 on consumer hardware?

Our compatibility engine estimates on the order of 12 tokens/sec on a mid-range GPU and up to 30 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.

Which quantization of Mistral Small 4 should I download first?

Start with Q4_K_M (72 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.

Tools

Is Mistral Small 4 the right pick for you?

Compute self-hosted ROI → Back to catalog