Mixtral 8x7B
By Mistral AI · France
Updated 2026-07-13
Overview
The Mistral AI MoE that popularized open-weight sparse models. Eight 7B experts deliver 47B-class output, but you pay 47B-class VRAM costs.
When to pick this model
- Inference servers with ample VRAM where speed-per-quality matters
- Workloads needing strong multilingual and coding performance
- Apache 2.0 commercial deployments
- Comparative benchmarks against newer dense models
- Fine-tuning research on a well-documented MoE architecture
VRAM requirements by quantization
| Quantization | VRAM required |
|---|---|
| Q4_K_M (recommended) | 26 GB |
| Q5_K_M | 32 GB |
| Q8_0 | 50 GB |
| FP16 (no quantization) | 94 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 8x7B spills past single consumer GPUs even at Q4_K_M (26 GB) — think dual-GPU or workstation cards. Stepping up to Q8_0 nearly doubles the footprint to 50 GB, and unquantized FP16 weights take 94 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.
Without a GPU, Mixtral 8x7B needs roughly 48 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 35 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.
What hardware do you need
The table below matches Mixtral 8x7B 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 8x7B |
|---|---|---|
| 8 GB | RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB | Does not fit — needs 26 GB at Q4_K_M |
| 12 GB | RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop | Does not fit — needs 26 GB at Q4_K_M |
| 16 GB | RTX 5080, RTX 4080 Super, Radeon RX 9070 XT | Does not fit — needs 26 GB at Q4_K_M |
| 24 GB | RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop | Does not fit — needs 26 GB at Q4_K_M |
| 32 GB | RTX 5090 | Q5_K_M (32 GB used) |
Which GPU should you buy to run Mixtral 8x7B?
To run Mixtral 8x7B locally at Q4, you need ~26 GB of VRAM. The best value for this is a RTX 5090 (32 GB VRAM).
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 | 70.6 |
| HumanEval | 40.2 |
| HellaSwag | 86.7 |
Scores published by the model author or aggregated from public leaderboards. Re-measured monthly by our editorial team.
To put Mixtral 8x7B in context: its MMLU score of 70.6 ranks #21 of the 34 catalog models with a published MMLU result (catalog median 73.4); its HumanEval score of 40.2 ranks #24 of the 26 catalog models with a published HumanEval result (catalog median 81.1). Scores are author-published and measured under different harnesses, so treat gaps of a few points as noise rather than a verdict.
Strengths
- Quality well above dense models of equivalent active params
- Strong coding and multilingual performance
- Apache 2.0 license
- Battle-tested in production stacks
Limitations
- Roughly 26GB VRAM at Q4 — same footprint as a dense 47B
- Eclipsed by Qwen 3 and Llama 3.3 in 2025 benchmarks
- 32k context now feels limiting
- Knowledge cutoff predates current tooling
Typical workloads
In our catalog grid, Mixtral 8x7B is filed under Pro Assistant, Long Analysis — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: everyday chat, drafting and summarization.
The 32k-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: MoE 8×7B · 32 experts, 2 active per token · 47B total / 13B active
Training: Mistral AI multilingual corpus. First popular large open-weight MoE.
A historically important MoE, now a second-tier choice — newer dense 24-32B models match it for less VRAM.
Quick start
ollama run mixtral:8x7bOr 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 8x7B need?
At the recommended Q4_K_M quantization, Mixtral 8x7B needs about 26 GB of VRAM. Q8_0 takes 50 GB, and unquantized FP16 weights take 94 GB.
Can Mixtral 8x7B run without a GPU?
Yes — with roughly 48 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 8x7B support?
Mixtral 8x7B supports a 32k-token context window (32,768 tokens).
Can I use Mixtral 8x7B commercially?
Yes. Mixtral 8x7B is released under Apache 2.0, a permissive open-source license that allows commercial use, modification and redistribution.
How fast is Mixtral 8x7B on consumer hardware?
Our compatibility engine estimates on the order of 12 tokens/sec on a mid-range GPU and up to 35 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.
Which quantization of Mixtral 8x7B should I download first?
Start with Q4_K_M (26 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.