Mixtral 8x22B.
Mistral's open MoE — 141B total, 39B active.
1× Intel Gaudi2.
Most-aggressive quantisation we have a working recommendation for. Lower precision = less VRAM = cheaper hardware, at a small accuracy cost.
Cheapest hosted endpoints.
What it's best at.
Workloads.
Frequently asked.
How do I run Mixtral 8x22B?
Where can I access Mixtral 8x22B?
How much does it cost to run Mixtral 8x22B?
Is Mixtral 8x22B open-source or proprietary?
Cheapest hardware per quantisation.
Each row is one quantisation tier (the same weights compressed differently). Lower precision → lower VRAM → cheaper hardware, at the cost of small accuracy loss. $/hr refreshed hourly from each provider's API.
| Quantisation | Cheapest GPU config | Total VRAM | Live $/hr | tokens/sec | |
|---|---|---|---|---|---|
|
FP16
FP16 — half precision (default)
|
384 GB | — | — | Compare → | |
|
FP8
FP8 — 8-bit float (Hopper / Blackwell)
|
192 GB | — | — | Compare → | |
|
INT4
INT4 — 4-bit integer (~4× VRAM saving)
|
96 GB | — | — | Compare → |
What it costs per month across providers.
Estimate your monthly bill for Mixtral 8x22B across every host that publishes per-token pricing. Slide your token volumes; the chart + table re-rank cheapest-first.
Cheapest provider on the left.
Total monthly cost — input + output tokens combined.
Bill breakdown.
Rent the GPU instead of paying per token.
For an open-weights model like Mixtral 8x22B, you can rent a GPU and serve inference yourself. The math: cheapest GPU rental × 730 hours/month + your electricity rate × power draw.
Assumes the GPU runs 24/7 at ~85% utilisation. If your traffic is bursty, you'll pay less for the API and probably more for the GPU (idle hours still cost rental). The breakeven analysis lives on the Self-host vs API breakeven tool.
What it's best at.
Scores normalised against benchmark ceilings (100 = perfect). Coloured by tier — coral 80+ frontier, lavender 65+ strong, sage 50+ solid, slate below.
Published scores.
| Benchmark | Score | Source |
|---|---|---|
| MATH | 41.8 | official ↗ |
| MMLU | 77.8 | official ↗ |
| HumanEval | 76.0 | official ↗ |
About Mixtral 8x22B.
Mixtral 8x22B is Mistral's open MoE (Mixture-of-Experts) model — 141B total parameters, but only 39B activated per token (top-2 routing across 8 experts). Apache-2.0 licensed, the most permissive of the Mistral family. Inference cost ~equivalent to a 39B dense model while quality approaches 70B-class dense models. Strong on coding (HumanEval 76%) and math. 64K context. Mostly superseded by Mistral Large 2 for new builds but still popular in cost-sensitive deployments because of the Apache license.