whatmodel.sh / Mixtral 8x22B v0.1 (base)
Mixtral 8x22B v0.1 (base)
↗ share
+ contribute
Mixtral 8x22B v0.1 (base) is an open-weights language model from Mistral, released April 2024. Its published size is 140.62B parameters (39B active per token) and a 66k-token context window. Licence: apache-2.0.
Catalogue identity notes
Maker repo mistralai/Mixtral-8x22B-v0.1 @e1cd34f: licence tag apache-2.0. 140,620,634,112 parameters from the Hugging Face safetensors metadata; active 39B per token (Mistral 'Cheaper, Better, Faster, Stronger' (Mixtral 8x22B): 'only 39B active parameters out of 141B'). Released 2024-04: month the maker's Hugging Face repo was created (createdAt 2024-04-16). Pretrained base checkpoint (not an instruct/chat version).
$ how it compares
The LLMs ranked either side of Mixtral 8x22B v0.1 (base), out of 1132 with a score, then more from Mistral.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 663 |  Qwen1.5 110B Chat | 47±1 | local | ||
| 664 |  WhiteRabbitNeo-33B-v1 | 47±1 | local | ||
| 665 | Mixtral 8x22B v0.1 (base) | 47±3 | local | ||
| 666 |  DeepSeek R1 Distill Llama 8B | 47±1 | local | ||
| 667 |  EXAONE 4.0 1.2B | 47±2 | local | ||
| more from Mistral | |||||
| 88 | Mistral Large 4 (public preview) | 71±2 | api $2.06/M | ||
| 204 | Mistral Medium 3.5 (128B) | 65±1 | api $3.00/M +local | ||
| 212 | Mistral Medium 3.1 (2508) | 64±2 | comparison only | ||
$ what it's good at
other tests1 not in a task · 96% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| BigCodeBench (Complete) | 45.5% ✎ | ▂▁▁▄▂▅▇█▆▂▂▂ beats 42/58 | 96% |
$ where to run it
Other ways to run it
- Run it as published · needs ~281.24 GB · from huggingface.co/mistralai
uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='mistralai/Mixtral-8x22B-v0.1', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"
Weights and documentation: huggingface.co/mistralai ↗
$ how the score was worked out · BigCodeBench paper (Zhuo et a… 96%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 BigCodeBench paper (Zhuo et al., ICLR 2025), arXiv 2406.15877v4 | 1 | 96% |
| prior · pull toward the middle | 4% | |
| weighted average | = 47.0 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json