Mixtral 8x22B Instruct
By Mistral AI · France
Updated 2026-07-13
Overview
Mistral AI's mature 141B/39B-active MoE under Apache 2.0, scoring 77.8 on MMLU and 45.1 on HumanEval. A proven general-purpose workhorse at roughly 80GB in Q4.
When to pick this model
- Stable, well-understood production deployments
- Apache-licensed commercial products
- Multilingual general chat including French
- Workloads where reliability beats latest benchmarks
- Teams with existing Mixtral infrastructure
VRAM requirements by quantization
| Quantization | VRAM required |
|---|---|
| Q4_K_M (recommended) | 82 GB |
| Q5_K_M | 100 GB |
| Q8_0 | 150 GB |
| FP16 (no quantization) | 282 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, Mixtral 8x22B Instruct is server-class even at Q4_K_M (82 GB). Stepping up to Q8_0 nearly doubles the footprint to 150 GB, and unquantized FP16 weights take 282 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.
Without a GPU, Mixtral 8x22B Instruct needs roughly 120 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 2 tokens/sec on entry-level GPUs, on the order of 8 tokens/sec on a mid-range card, and up to 22 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.
What hardware do you need
The table below matches Mixtral 8x22B Instruct 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 Mixtral 8x22B Instruct |
|---|---|---|
| 8 GB | RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB | Does not fit — needs 82 GB at Q4_K_M |
| 12 GB | RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop | Does not fit — needs 82 GB at Q4_K_M |
| 16 GB | RTX 5080, RTX 4080 Super, Radeon RX 9070 XT | Does not fit — needs 82 GB at Q4_K_M |
| 24 GB | RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop | Does not fit — needs 82 GB at Q4_K_M |
| 32 GB | RTX 5090 | Does not fit — needs 82 GB at Q4_K_M |
Which GPU should you buy to run Mixtral 8x22B Instruct?
To run Mixtral 8x22B Instruct locally at Q4, you need ~82 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.
Published benchmark scores
| Benchmark | Score |
|---|---|
| MMLU | 77.8 |
| GSM8K | 78.6 |
Scores published by the model author or aggregated from public leaderboards. Re-measured monthly by our editorial team.
To put Mixtral 8x22B Instruct in context: its MMLU score of 77.8 ranks #12 of the 34 catalog models with a published MMLU result (catalog median 73.4); its GSM8K score of 78.6 ranks #8 of the 9 catalog models with a published GSM8K result (catalog median 83.1). Scores are author-published and measured under different harnesses, so treat gaps of a few points as noise rather than a verdict.
Strengths
- Battle-tested mature MoE
- Strong general-purpose performance
- Apache 2.0 license
- Solid multilingual coverage
Limitations
- 80GB in Q4 still demands serious hardware
- Coding trails newer specialists
- 64K context lags 2026 competitors
- Outclassed by newer Mistral releases on most benchmarks
Typical workloads
In our catalog grid, Mixtral 8x22B Instruct is filed under Pro Chat, Multilingual, Agents — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: multilingual workloads; French-language output where quality matters.
The 62k-token context window covers long chats and mid-sized documents, though very large retrieval workloads will need chunking. The Apache 2.0 license is permissive, so shipping it inside a commercial product raises no special legal questions.
Architecture & training
Architecture: Sparse MoE · 8 experts · 141B/39B active · GQA
Training: Apache 2.0, 64k ctx.
Still a dependable Apache-licensed generalist, but newer Mistral models now beat it across the board.
Quick start
ollama run mixtral:8x22bOr 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 Mixtral 8x22B Instruct need?
At the recommended Q4_K_M quantization, Mixtral 8x22B Instruct needs about 82 GB of VRAM. Q8_0 takes 150 GB, and unquantized FP16 weights take 282 GB.
Can Mixtral 8x22B Instruct run without a GPU?
Yes — with roughly 120 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 Mixtral 8x22B Instruct support?
Mixtral 8x22B Instruct supports a 62k-token context window (64,000 tokens).
Can I use Mixtral 8x22B Instruct commercially?
Yes. Mixtral 8x22B Instruct is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.
How fast is Mixtral 8x22B Instruct on consumer hardware?
Our compatibility engine estimates on the order of 8 tokens/sec on a mid-range GPU and up to 22 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.
Which quantization of Mixtral 8x22B Instruct should I download first?
Start with Q4_K_M (82 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.