Mixtral 8x22B Instruct
Mistral AI · released Apr 17, 2024
Mistral's official instruct fine-tuned version of [Mixtral 8x22B](/models/mistralai/mixtral-8x22b). It uses 39B active parameters out of 141B, offering unparalleled cost efficiency for its size. Its strengths include: - strong math, coding,...
Specification
- Context window
- 66K
- Max output
- 52K
- Knowledge cutoff
- Jan 31, 2024
- Parameters
- 140.6B
- Licence
- apache-2.0
- Serving providers
- 1
- Moderated
- No
Intelligence
5.7
11th percentile
Coding
—
Coding Index
Agentic
—
Agentic Index
Output speed
103 t/s
Median across providers
Latency
499ms
Time to first token
Cost per task
$0.25
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- MMLU-Pro53.7%
- GPQA Diamond33.2%
- LiveCodeBench14.8%
- Humanity's Last Exam4.0%
- AIME 20250.0%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| MMLU-Pro | 53.7% |
| GPQA Diamond | 33.2% |
| LiveCodeBench | 14.8% |
| Humanity's Last Exam | 4.0% |
| AIME 2025 | 0.0% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- Granite 4.1 8BIBM Granite6.4
- Olmo 3 32B ThinkAllen AI6.1
- Hermes 3 70B InstructNous Research6.0
- Phi 4Microsoft5.9
- Mistral LargeMistral AI5.8
- Mixtral 8x22B InstructMistral AI5.7
- Reka Flash 3Reka AI5.6
- GPT-3.5 TurboOpenAI5.5
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Granite 4.1 8B | 6.4 |
| Olmo 3 32B Think | 6.1 |
| Hermes 3 70B Instruct | 6.0 |
| Phi 4 | 5.9 |
| Mistral Large | 5.8 |
| Mixtral 8x22B Instruct | 5.7 |
| Reka Flash 3 | 5.6 |
| GPT-3.5 Turbo | 5.5 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $2
- Output / 1M tokens
- $6
- Cached input / 1M
- $0.2
- Blended 3:1
- $3
One task, estimated
$0.25
- Input tokens
- 50,000
- Output tokens
- 25,000
- Profile
- Standard
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.