VP Science @mistralai. Ex RS @metaAI | PhD @ENSdeLyon | MVA @ENS_ParisSaclay
Paris, France
Joined July 2010
- Tweets191
- Following275
- Followers2.7K
- Likes1.2K
ML4 is now open-weight! Frontier capabilities on code agents & cyber, way more efficient and compact than its Preview version three weeks ago (!) and cherry on the cake: technical report telling its epic backstory. Check it out: huggingface.co/mistralai/Mis…
Good stuff, chonky stuff!
We just released an updated version of le Chonk (ML4) on the API and on HuggingFace. It is simply the best OSS model for code!
Le Chonk pushes the OSS Pareto frontier for the AA coding agent index: tied with GLM5.3 using close to half of the tokens.
It is the best OSS model on FrontierCode 2.0, which came out earlier this week, and on Terminal-Bench 4.0! 🐈
Very excited to release Voxtral 2!
- Voxtral Realtime: a SOTA real-time transcription model under an Apache 2 license, with latency configurable to sub-200 ms
- Voxtral Mini Transcribe 2: a SOTA transcription model with diarization, word-level timestamps, and context biasing
Ministral 3B and 8B, your perfect models for edge/agentic use-cases!
We just released two small models, with 3B and 8B parameters. Ministral 3B is exceptionally strong, outperforming Llama 3 8B and our previous Mistral 7B on instruction following benchmarks.
mistral.ai/news/ministraux/
First multimodal model, a 12B, on par with the other big ones. And we’re not done cooking!
Exciting update from Vision Chatbot Arena!
We’ve gathered over 6K new votes for the latest open vision models (Qwen2, Llama 3.2, Pixtral) and ChatGPT-4o.
- ChatGPT-4o has taken the #1 spot, surpassing Gemini.
- Open models (Qwen, Llama 3.2, Pixtral) are rapidly improving, matching proprietary offerings
Competition in vision is heating up. Cast your vote and help decide the best vision model!
Full leaderboard link below👇
![[Deprecated] Pixtral 12B | Mistral AI](https://pbs.twimg.com/card_img/2120171092346343424/v4a2b1P0?format=jpg&name=800x419)
Thrilled to see start-ups like Pretzel AI build products on top of our code model @MistralAI, dubbed Codestral! github.com/pretzelai/pretzel…
My Tailor is Mistral!
Model Customization API:
- Finetune a Mistral model efficiently with LoRA in a few clicks
- Get a cost-effective personnalized API endpoint
It's been (exactly) a year since we started and it's only the beginning!
Blog post: lnkd.in/gFQnUERB
Super excited to release Codestral 22B, our first-ever code model @MistralAI, capable of understanding 80+ programming languages. Codestral is way smaller and faster than its competitors while performing similarly or even better! Blogpost: lnkd.in/eGK66TMN
You can try it now on Le Chat to generate code given specific instructions, explain a code snippet, write documentation or factorize convoluted functions: chat.mistral.ai/chat
1
1
365
Cheaper, Better, Faster, Stronger!
Mixtral 8x22B Instruct by @MistralAI is out!
- Apache 2.0
- 64k sequence length
- Built-in function calling and multilingual!
Outperforms all open models, cheap to serve, available at: huggingface.co/mistralai/Mix…
Blog: mistral.ai/news/mixtral-8x22…
Extremely excited by our new release @MistralAI:
- Mistral Large: our new flagship model: mistral.ai/news/mistral-larg…
- Le Chat: a chat interface for our model family chat.mistral.ai/
- Function calling, JSON mode, new very efficient small endpoint and way more!
Mixtral 8x7B is here, 11 weeks only after Mistral 7B. Outperforms Llama 2 70B and GPT 3.5 on most benchmarks, at the inference cost of a 12B dense model, with 32k tokens context size.
Also, check out our open positions in Paris & London!
We will be at ICCV in Paris for the next couple of days. Feel free to reach out if you are interested about @MistralAI !
7
608
Super proud to release our very first model, the best 7B in town, after an intense summer at @MistralAI!
Mistral 7B is out. It outperforms Llama 2 13B on every benchmark we tried. It is also superior to LLaMA 1 34B in code, math, and reasoning, and is released under the Apache 2.0 licence.
mistral.ai/news/announcing-m…
On my way to #NeurIPS22, can’t wait to talk about Privacy-Preserving ML and Federated Learning with a zest of efficiency. Let’s connect!
8
We are hosting an Masters internship at Meta AI in the field of Differential Privacy with @alexsablay in summer 22' that could result on a CIFRE PhD. Reach out quickly if you're interested!
We are hosting a research internship in Efficient Federated Learning with @alexsablay at Facebook AI Paris/London next summer (2022). If you are currently pursuing a PhD and are interested in these topics, let’s get in touch!