Super predictable.

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!

Congrats to the Openrouter team on the release. However stop using TTFT! In practice it randomly penalizes providers who get jobs with larger prefill. Instead if a users product cares about low latency, measure Queue time, Prefill tok/s, Decode tok/s. Then match the provider based on the users traffic shape.

Hot take: The compute credit crunch is not real and Elon is fearmongering. I see no slowdown in my non-IG pipeline (deals under $500M mostly) Real issues: - Higher interest rates - Supermicro delaying all their deliveries

I have talked to security teams from small to fortune500 companies in the past week. It’s deeply worrying that <1% of them are using abliterated models. These models are a step function in offensive capability that adversaries now have access to. The bottle is open and the only way is forward and we need to move fast as possible. Skeptical? Happy to give you $10 to try them on @BoundlessHQ. Access below.

Attackers already use models that don’t refuse, and defenders should have the same. Boundless Defensive Access gives security teams DeepSeek V4.1 Flash 1017 Abliterated, a version with its refusal behavior removed. Apply for access: boundless.network/defensive-…

Boundless is top 3 for throughput on Vercel AI Gateway. 367 tokens per second on DeepSeek V4.1 Flash 1017. Fast inference starts with Boundless.

In a single year, AI has rewritten the security priority list for CISOs. Attackers are moving faster. Bain estimates AI compresses a typical cyberattack from about four weeks to about 18 hours, and models keep getting better at finding and exploiting weaknesses. There's also more to protect, with employees shipping apps they couldn't have built a year ago and agents getting access to source code, documents, and production systems. In a survey of 100 CISOs at companies with 1,000+ employees, all seven of the most-selected 2027 priorities involve AI. The biggest jumps since 2026: - Internal AI agent security (+28 pts) - AI SOC (+26 pts) - AI offensive security (+24 pts) - AI code security (+23 pts) - AI vulnerability remediation (+22 pts) Last year's staples: - Exposure & vulnerability management (-21 pts) - SIEM / security data (-18 pts) - Privileged access management (-18 pts) From a16z's @OwenColby and @steph_zhang on The New Security Agenda: a16z.news/p/the-new-security…

oh? @BoundlessHQ

Got so frustrated with how outdated dependabot was I just built my own on top of Boundless instead

Are we entering another “Google is back” cycle? Google’s new model Gemini 4 Carbon beats Opus 5.5 and GPT-6 Astra on most benchmarks and ranks #1 on the Artificial Analysis Intelligence Index v5. Let’s see if it can shake up Anthropic’s IPO.

For Halloween, I've been doing some brain surgery on an open model 🎃🧟 Taking inspiration from @_lyraaaa_ 's work Push 2 of its neurons and Qwen3.5-4B becomes a cookie lover. Push 1 and it's a lawyer Built a small app on top of @BoundlessHQ so anyone can try 👇

For Halloween, I've been doing some brain surgery on an open model 🎃🧟 Taking inspiration from @_lyraaaa_ 's work Push 2 of its neurons and Qwen3.5-4B becomes a cookie lover. Push 1 and it's a lawyer Built a small app on top of @BoundlessHQ so anyone can try 👇

Big updates to our Alexa product lineup today. I can’t wait for you to experience these. Pumped might even be an understatement.

How are companies decreasing AI costs? Shiv Shankar (@sshankar) breaks down the AI adoption lifecycle at Fortune 500 companies, and why they all end up on open source models that cut costs by 95%. Full episode below.

What is an open weight model? Shiv Shanker (@sshankar), founder and CEO of Boundless, joins the Dry Run to explain what an open weight model is, and why he thinks most companies will try closed models first but eventually shift to open ones. 00:40 Understanding open weight models 01:57 What labs actually share when they publish weights 06:33 Cost considerations in AI adoption 08:47 Security and data privacy 13:08 You can't vibe code DocuSign 15:48 The MySQL playbook for open source profits 20:33 Why the top open weight models come from China 24:09 What changes after OpenAI and Anthropic go public 25:32 The case against open weight models 30:04 Don't give an agent your passwords 31:50 Regulatory frameworks for open weight models 36:23 Education and workforce development for AI 38:41 The sign that open weight models are winning

Will the majority of tokens flow through open or closed models? “My suspicion is that there will always be a market for frontier. There will always be some new thing. Science may be the next thing that frontier is being set at. But I think predominantly token usage will be commodity token usage.” @satyanadella Love to hear your thoughts @natolambert @lqiao @Thom_Wolf @swyx