whatmodel.sh / Mistral Large 4 (public preview)
Mistral Large 4 (public preview)
↗ share
+ contribute
Mistral Large 4 (public preview) is a closed-weights language model from Mistral, released October 2026. Its published size is 1050B parameters (52B active per token) and a 1M-token context window. Licence: proprietary.
Catalogue identity notes
Mistral AI's Mistral Large 4, API model 'mistral-large-4' (alias 'mistral-large-4-0'), version 26.10, launched 6 October 2026 as a 'Public Preview' that anyone can call on Mistral's API ('You can try the preview API today on Mistral Studio'). Mistral describes it as open-weight, but the weights are not out yet: the model page's Weights link (huggingface.co/mistralai/Mistral-Large-4-1T-A52B) says 'Coming soon', the repository returns HTTP 401, and the announcement says 'Weights drop end of this month'. It is therefore recorded as open_weights false with no local variant and licence 'proprietary' for now; switch both when the weights and their licence are published. Arena entry 'mistral-large-4' (organization mistral, licence 'Proprietary') first appears in the 2026-10-08 text, webdev and agent snapshots, after the public API launch. A new model, not Mistral Large 3 (mistral-large-2512).
Model information sources
Release date ↗ · Parameter count ↗
Mistral's model page: '52B active parameters and 1.05T total parameters, and a 1.6B vision encoder'. Observed 2026-10-10.
$ how it compares
The LLMs ranked either side of Mistral Large 4 (public preview), out of 828 with a score, then more from Mistral.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 77 |  GPT-5.4 mini | 71±1 | api $1.69/M | ||
| 78 |  Ling 3.0 Flash (124B-A5B) | 71±2 | api $0.090/M +local | ||
| 79 | Mistral Large 4 (public preview) | 71±2 | api $2.06/M | ||
| 80 | GPT-5.1 | 71±1 | api $3.44/M | ||
| 81 |  GLM-5.1 | 71±1 | api $2.15/M +local | ||
| more from Mistral | |||||
| 174 | Mistral Medium 3.5 (128B) | 65±1 | api $3.00/M +local | ||
| 180 | Mistral Medium 3.1 (2508) | 64±2 | comparison only | ||
| 200 | Magistral Medium 1.2 (2509) | 63±2 | comparison only | ||
$ what it's good at
agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #62 of 300 * · 38% of its score · 1 test
Beats Gemini 3.6 Flash, GPT-5 Pro, LongCat 2.0 (1.6T-A48B) here, though they rank higher overall. Trails MiniMax-M3, GLM-5.1 here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena Agent (October 2026 aggregate IPS) | -0.0656 IPS ✎ | ▁▁▁▄▂▃▇█▂▂▁▃ beats 9/49 | 38% |
coding▁▁▁▃▆██▇▆▄▃▁ #39 of 458 * · 34% of its score · 1 test
Beats GPT-5.5, GPT-5.6 Terra, GPT-5.4 here, though they rank higher overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena WebDev (Code Arena) | 1541.3 ✎ | ▁▂▄▃▅█▄▇▆▄▁▁ beats 82/119 | 34% |
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #79 of 393 * · 19% of its score · 1 test
Beats LongCat 2.0 (1.6T-A48B) here, though they rank higher overall. Trails Gemma 4 12B, Gemma 4 26B-A4B, Nemotron Cascade 2 30B-A3B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| LMArena (text) | 1429.3 ✎ | ▁▁▁▂▃▃▃▆▆▇█▃ beats 283/379 | 19% |
other tests1 not in a task · 7.3% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| SimpleQA Verified | 20.0% ✎ | ▁▁▃▂▃▃▆▆█▃▂▅ beats 12/78 | 7.3% |
$ where to run it
Through an API
api · Mistral AI · $2.06/M · $1.36 in / $4.18 out per 1M tokens · 1M context · docs.mistral.ai ↗ · price sourceObserved 2026-10-10 on Mistral's own Mistral Large 4 model page (Price, USD): original (list) price $1.36 input / $4.18 output per million tokens; cached input $0.14. The page also shows a sale price of $0.68 input / $0.07 cached / $2.09 output, which is not used here. API names mistral-large-4 and mistral-large-4-0, version 26.10, status 'Public Preview'.
$ how the score was worked out · LMArena leaderboard dataset 92%, Epoch AI — internal evaluatio… 7%
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 |
|---|---|---|
 LMArena leaderboard dataset | 3 | 92% |
 Epoch AI — internal evaluations | 1 | 7% |
| prior · pull toward the middle | 1% | |
| weighted average | = 71.1 |
$ 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