majoub 🦇🔊 retweeted
La Chine avertit que si Israël utilise une arme nucléaire contre un quelconque pays, y compris l'Iran.
Ce sera la fin d'Israël.
🚨 LA SITUATION ENTRE LES ÉTATS-UNIS 🇺🇸 ET L’IRAN 🇮🇷 EST OFFICIELLEMENT TRÈS INQUIÉTANTE !
La télévision d’État iranienne diffuse désormais des consignes de sécurité à la population en cas d’éventuelle attaque nucléaire ☢
▫️Ne pas regarder la lumière de l'explosion
▫️Entrer dans le bâtiment le plus proche et s'éloigner des fenêtres et des murs extérieurs
▫️Retirer les vêtements portés à l'extérieur et les placer dans un sac fermé
▫️Rester à l'intérieur pendant au moins 24 heures
majoub 🦇🔊 retweeted
What we’ve been working on
Since our last update on October 8th, we’ve been focused on improving the infrastructure behind robinhood:0x56910d4409f3a0c78c64dd8d0545ff0705389870.
Distributions were paused from October 18th to the 29th while the system was rebuilt. Protocol fees continued to accrue throughout that period and were deployed into stocks once the new setup went live.
Since October 29th, three distribution cycles have completed, sending roughly $23,600 in stocks to 4,163 wallets.
The upgraded setup includes:
• Fixed 36-hour cycles for a consistent distribution schedule.
• Former holders made whole, with around $3,680 in banked stocks distributed to 5,366 addresses that had fallen below the 10,000 robinhood:0x56910d4409f3a0c78c64dd8d0545ff0705389870 threshold.
• Built-in price protection, with stock purchases and ETH → USDG conversions limited to within 1.5% of trusted prices.
The result is a more predictable, secure and scalable distribution system.
Closing the volume gap
A significant portion of robinhood:0x56910d4409f3a0c78c64dd8d0545ff0705389870 trading volume currently routes through pools that do not use the official fee hook, meaning those trades do not contribute to the 3% fee funding stock distributions.
We have already addressed one side of this issue.
Pools holding robinhood:0x56910d4409f3a0c78c64dd8d0545ff0705389870 were previously able to appear as holders and receive distributions. Those pools are now excluded, receive $0, and new pools are checked before every cycle.
The remaining fee-bypass issue requires a deeper change and is already in development. We’ll have more to share on the full approach soon.
The objective is simple: robinhood:0x56910d4409f3a0c78c64dd8d0545ff0705389870 trading activity should contribute to the protocol economics it was designed to support.
Expanding the Index ecosystem
A large part of our recent product work has also gone into Blend.
Blend introduces permissionless onchain Baskets backed by tokenized assets. Each Basket represents defined quantities of its underlying assets in a single share and can be bought, sold, redeemed and integrated into other applications.
For robinhood:0x56910d4409f3a0c78c64dd8d0545ff0705389870, Blend adds another source of protocol activity and revenue, expanding the ways onchain activity can ultimately flow back into stocks.
What comes next
The distribution infrastructure is stronger, the fee-routing issue is being addressed, and Blend is expanding the ways tokenized assets can be used across the Robinhood ecosystem.
Our focus remains the same: build useful infrastructure around onchain stocks and return more of the value generated by that infrastructure back into the Index ecosystem.
Stocks for everyone.
majoub 🦇🔊 retweeted
Three weeks of training camp for Le Chonk. No days off.
Thanks for putting it to the test, @ThibaultJaigu and @RequestyAI.
Impressed by the new @MistralAI large-4 update! The model update is much better at coding and performs as the 2nd best OSS model on our coding benchmark
majoub 🦇🔊 retweeted
Mistral Large 4 is now open weights, narrowly making it the #1 open weights model developed outside of China
@MistralAI has updated and released the weights of their flagship model, Mistral Large 4, with 1.05T total and 52B active parameters. With high reasoning, Mistral Large 4 scores 26 on the Artificial Analysis Intelligence Index v5, up 1 point from Mistral Large 4 Preview (25), with the major differences being increased token efficiency and reduced hallucinations. It sits 1 point ahead of the leading US-developed open weights model, Muse Spark Open (xhigh, 25).
On the Artificial Analysis Cyber Index, Mistral Large 4 (high) scores 51, third among current open weights models behind DeepSeek V4.1 Flash 1017 (max, 57) and MiMo-V2.6-Pro (56).
Key takeaways:
➤ Narrowly leads non-Chinese open weights models: Mistral Large 4 (high) leads Muse Spark Open (xhigh), the next-best open weights model developed outside China, with the largest gap on Terminal-Bench 4.0 (25% vs. 8%). It trails leading open weights models DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) by 6 points on the Intelligence Index, largely due to weaker coding and agentic knowledge work performance: 15% on FrontierCode 2.0 (vs. 29% and 22%, respectively) and 1349 Elo on AA-Briefcase (vs. 1570 and 1504)
➤ Increased token efficiency: Mistral Large 4 (high) uses 82K output tokens per Intelligence Index task, 39% fewer than Preview (135K), placing it on the open weights Pareto frontier. The seven open weights models that score higher use between 113K and 225K output tokens per task
➤ #2 open weights model on the Coding Agent Index: On the Artificial Analysis Coding Agent Index, Mistral Large 4 (high) scores 48 in Mistral Vibe while using 8.5M tokens per task, about half as many as the leading open weights model GLM-5.3 (max) in Opencode (49, 15.9M). It also scores above another leading open weights model, DeepSeek V4.1 Flash 1017 (max), in DeepSeek Harness (46, 16.7M)
➤ Reduced hallucinations: Mistral Large 4 (high) has a 28% hallucination rate on the updated AA-Omniscience v1.1, down from 41% for Preview and matching Muse Spark Open (xhigh). However, its accuracy remains relatively low at 26%, similar to Ling 3.1 Flash 1019 (27%), and behind DeepSeek V4.1 Flash 1017 (max, 40%) and Muse Spark Open (xhigh, 42%)
➤ Lower price and cost per task than the Preview: Mistral Large 4 is priced at $0.70/$2.10 per 1M input/output tokens, a ~50% reduction from the Preview's pricing of $1.36/$4.18. At the new pricing, it costs $2.06 per Intelligence Index task, down 63% from $5.54 for the Preview. This is comparable to Muse Spark Open (xhigh, $2.02) but more costly than leading open weights models DeepSeek V4.1 Flash 1017 (max, $0.60) and MiMo-V2.6-Pro ($0.26)
Key model details:
➤ Size: 1.05T total parameters / 52B active parameters
➤ Context window: 1M tokens
➤ Multimodality: Text and image input, text output
➤ Pricing: $0.70/$2.10 per 1M input/output tokens ($0.07 per 1M cached input tokens)
➤ License: MEOW License (Mistral Ecosystem Open-Weight License), where operating a Model-as-a-Service business requires a separate commercial license with Mistral
Congratulations to the team at @MistralAI on the release!
majoub 🦇🔊 retweeted
Congrats @MistralAI team on the next gen Le Chonk! 🙌
1T-parameter MoE multimodal model trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs.
Our day 0 support:
NVFP4 on @huggingface → huggingface.co/mistralai/Mis…
Scale distributed inference using performance-optimized Dynamo → docs.nvidia.com/dynamo/recip…
Today, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity.
💻 Built to build: Le Chonk ranks #1 on @cognition’s FrontierCode among open models, and sets a new benchmark for Pareto efficiency on @artificialanalysis.
🛡 Built to defend: State-of-the-art in cyberdefense, put through the paces by the world's leading cyber vendors, including @CrowdStrike, @cloudflare, @rubrik, @zscaler, and others.
Built for sovereignty: Make it yours. Download the weights on @huggingface , customize it with Mistral Forge, and deploy it wherever you need it.
Read more from our team and customers 👇:
majoub 🦇🔊 retweeted
Today, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity.
💻 Built to build: Le Chonk ranks #1 on @cognition’s FrontierCode among open models, and sets a new benchmark for Pareto efficiency on @artificialanalysis.
🛡 Built to defend: State-of-the-art in cyberdefense, put through the paces by the world's leading cyber vendors, including @CrowdStrike, @cloudflare, @rubrik, @zscaler, and others.
Built for sovereignty: Make it yours. Download the weights on @huggingface , customize it with Mistral Forge, and deploy it wherever you need it.
Read more from our team and customers 👇:
majoub 🦇🔊 retweeted
There isn't much to do onchain now: BTC is chopping, yield farms aren't worth it (R:R wise vs Tbills) and trenches are dead.
Don't burn money farming airdrops or playing with perps. You have no edge there. Zoom out from the game, don't do anything silly while everyone else PvP.
Instead, stay liquid.
And to kill time: read books, do sport, play with ai and enjoy life.
majoub 🦇🔊 retweeted
Désolé, mais cet état d'esprit des politiciens est idiot, et ça continuer de faire couler notre écosystème.
NON Mistral n'est pas la France et la France n'est pas Mistral.
Mistral est bien sûr une excellente boîte française, et c'est une chance pour la France de l'avoir.
Mais mettre tous les jetons du gouvernement sur un seul acteur, c'est condamner le reste de l'écosystème à ne pas innover.
A chaque fois que l'Etat veut choisir un champion, ça se finit mal. Simplement parce que, désolé monsieur Attal, mais vous êtes politicien et pas entrepreneur, vous n'avez aucune compétence pour savoir qui soutenir.
[Notons que le diagnostic serait différent sur des tâches d'infra, d'une ampleur telles que seul l'Etat peut les accomplir - c'est seulement là que l'Etat doit jouer son rôle]
Le rôle de l'Etat n'est pas d'attraper un joueur sur le terrain pour le pousser vers l'en-but. C'est plutôt de rendre le terrain praticable, en commençant par dégager les monceaux de lois qui nous empêchent de travailler.
Plutôt que d'aller serrer des mains pour avoir l'air dynamique, vous feriez donc mieux de consacrer ce temps à réfléchir : comment faire émerger d'avantages d'autres boîtes d'IA, EN PLUS de Mistral ?
Ça pourrait commencer par construire des datacenter, ou au moins débloquer leur construction par le privé : ça, c'est le rôle de l'Etat.
Mistral, c’est la France.
La France, c’est Mistral. 🇫🇷
Dans cette course à l’innovation et aux technologies, il y aura des perdants et des gagnants.
Je veux me battre pour que la France soit libre. Libre de développer ses propres modèles et outils.
@arthurmensch et ses équipes sont en première ligne dans ce combat. Ils font notre fierté.
Merci à lui pour son accueil aujourd’hui au siège de @MistralAI. Nous avons échangé longuement sur l’avenir de l’IA en France.
Avec une conviction : si nous faisons les bons choix, nous avons encore nos plus belles pages à écrire ! 🇫🇷 🇪🇺
majoub 🦇🔊 retweeted
Mistral, c’est la France.
La France, c’est Mistral. 🇫🇷
Dans cette course à l’innovation et aux technologies, il y aura des perdants et des gagnants.
Je veux me battre pour que la France soit libre. Libre de développer ses propres modèles et outils.
@arthurmensch et ses équipes sont en première ligne dans ce combat. Ils font notre fierté.
Merci à lui pour son accueil aujourd’hui au siège de @MistralAI. Nous avons échangé longuement sur l’avenir de l’IA en France.
Avec une conviction : si nous faisons les bons choix, nous avons encore nos plus belles pages à écrire ! 🇫🇷 🇪🇺
majoub 🦇🔊 retweeted
Today, we’re introducing the Responses API in public preview, made for building agents on Mistral.
Each response returns the model’s answer, its reasoning and its tool calls as separate items, so your app can handle every step more easily than with Chat Completions.
It works with all chat models hosted by Mistral, including Large 4.
