All entries

Our Black Hat talk on the OpenAI-Hugging Face incident is now live on youtube. This is a watershed moment for the industry. I encourage all defenders to watch, consider how attack dynamics will immine

Outlook where frontier AI is headed next 18 months: The AI reasoning training + harness loop works if you can produce enough data and reasoning traces (via verifiers). Proven with code and math result

OpenAI announced that Plus and Pro subscribers now have a slider to adjust the amount of reasoning effort ChatGPT applies to each response. The update employs GPT‑5.6 Sol for both Instant and deep rea

Plus and Pro users can access the updated version of GPT‑5.6 Sol in ChatGPT along with the new slider starting today. This version of GPT-5.6 Sol is for everyday chats, so it will only be available in

RIP to every 'which framework should I use' thread. harness on top, framework in the middle, runtime at the floor. one stack, three heights, fully composable. paste these four images into claude and a

Taalas buried the lede for the amazing demo of their first tape out. 15k tokens per second with a llama 8b model, ability to scale that up etched onto silicon As models satisfice etching makes sense,

The new GPT-5.6 Sol powers all chats for paid users, including Instant, creating one consistent experience. In our high-stakes factuality evaluation covering finance, medicine and law, the new GPT‑5.6

the smartest young people in sf were working on agi/alignment 5-10y ago. they are working on bci today. naomi is a total star and is one of an explosion of young talent into the bci space recently. i’

This definitely seems like something worth noting, and illustrates the gap between Fable/Astra class models and the previous frontier that was 'merely' good at hacking under human instructions. Initia

This paper by researchers from MIT and Stanford finds that most people would be financially better off if they followed the financial advice of LLMs (GPT-5.2 & Gemini 3 Flash) But some people get a bi

This webinar is happening in 30 minutes, and that means there's still time to register! Following the discussion will be an open Q&A with our Head of AI Education @Prof_OZ, and the @arizeai team. See

Together Serverless Inference gives developers a managed production path for bringing FLUX 3 into creative products and automated media workflows. Start building: https://www.together.ai/models/flux-3

Uber burned through its 2026 AI coding budget in four months. Microsoft canceled most of its Claude Code licenses six months after rolling them out. The mechanics are simple: per-token cost keeps fall

We made an MCP for your phone. Your laptop is just half of your life, and the other half is in your phone. Now your agent gets the mobile screen too. No connectors, no complex setup, it has access to

We re-tested GPT-5.6 Luna from @OpenAI on ARC-AGI (Verified) following its recent 80% price reduction: - ARC-AGI-2: 59.6%, 0.18/task - ARC-AGI-1: 90.7%, 0.07/task The new results match Luna's original

We're expanding our partnership with @huggingface to accelerate open science. Our storage on the Hub is roughly tripling to ~2 petabytes, & our downloads now run at high speed—even for our largest dat

We’re making better intelligence easier to access in ChatGPT for everyone: - GPT-5.6 Sol now powers both Instant and deep reasoning for Plus & Pro users, delivering more factual, focused responses. -

Yesterday, my OpenAI collaborator and I gave a detailed talk on the Huggingface incident, our models creating 'the message board', model misalignment, and more. https://www.youtube.com/watch?v=87DyyMV

99.9% uptime changes what your inference architecture has to survive. At Together AI, it means multi-data-center deployment, live traffic across both facilities, and enough capacity to absorb a full d

After rigorous testing, our joint AI project with Daiwa Securities is entering the full-scale production phase. We're bringing our agentic AI systems to @Daiwa_JP’s wealth management teams to accelera

Also I think AISI is a great model of a government agency tasked with AI security. They have open benchmarks, very fast testing, and clear communication about incidents that is neither hyped up nor hi

Announcing Discovery Loop! I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoL

Another real-world manifestation of the kind of misaligned actions frontier systems developed by leading companies can take to achieve goals. Current frontier AI models are trained with reinforcement

Big new release of my LLM CLI tool and Python library for talking to hundreds of different LLMs - reasoning traces, OpenAI Responses support, server-side tools, smarter logging and a whole lot more ht

ChatGPT Work is OpenAI's fighter in the highest stake product category in history: bringing the power of coding agents to the masses. It's also how a billion users will soon use ChatGPT by default. I

⚠️⚠️⚠️ Constitutional AI is not working. Aligning LLMs is not working. We need a different approach. If society doesn’t place its bets differently, we are screwed. Anthropic's Mythos created fake iden

DeepSeek V4 Flash 0731 is ready to fine-tune on Fireworks. Run SFT, DPO, and RL training jobs on the Dedicated Training API. SFT and DPO from the managed UI. Built for coding agents and high-volume pr

DeepSeek V4 Flash is Ollama's fastest growing model ever in token usage, and the most popular model on OpenRouter this week. It’s available in Pi across a range of providers. If you’ve never tried an

Document OCR is not Getting Commoditized (by Frontier Models) The most common question I get is whether frontier models are going to eat all document processing solutions - just screenshot the page an

Feels like Google could have been the dominating force in AI by open-sourcing the frontier with Gemini, Veo, and Nano Banana. Instead, they kept them behind APIs for a few billion dollars in revenue.

Fields Medalist Jacob Tsimerman knows MUCH MUCH more about math than I do, or ever will, but I have been studying natural and artificial intelligence for 40 years, and I think his prediction here (“AI

... four years later, I got the new Claude Fable 5 to actually build the game https://x.com/simonw/status/2085089518223602058 Four years ago today I tweeted about having GPT-3 and DALL-E come up with

Harness choice is a big deal. So much room to advance and improve results across the board with agent harnesses. Great paper highlighting this. New research releases DataSpace, a benchmark where data

Hey everyone! I just released a sub-6B sparse activation AI model which was built with a brand new architecture : fusion. I fused weights from @liquidai's LFM2.5-2.6B & @Alibaba_Qwen's Qwen3.6-35B-A3B

if you have been following his excellent work, @shloked has been breaking down every frontier labs' harness engineering for the last few months. excited to publish his deepest dive into ChatGPT yet as

Gary Marcus tweeted “Indeed.” and then Frank Rundatz replied that the deterministic harness required to ground a large‑language‑model agent differs by task. Because of this variability, Rundatz agrees

interesting (and completely opposed to @haider1’s take on the same graph) *the labs themselves* have only modestly changed their predictions on AGI timelines over the last decade. note also that (on a

The post contains two distinct snippets. First, user @GaryMarcus asks whether there is a more polite way to refer to “circle jerk.” Second, it shares a (likely satirical) claim that Microsoft’s AI rev

Just had to create an 'accidental-cyberattacks' tag on my blog We're up to four now: the original OpenAI+Hugging Face one, Anthropic's me-too attacks, then two new ones from the UK AI Safety Institute

.@Kimi_Moonshot benchmarked K3 endpoints across major inference providers. Together AI leads or ties for #1 on 3 of 4 benchmarks: OCRBench, MMMU Pro Vision, and DeepSWE. Open models like Kimi K3 have

Let's create with Qwen-Image-3.0-Pro on fal! 🎨 Qwen Image 3.0 Pro is now available on fal - Renders complex, dense typography - Preserves key details like facial features and identity while applying c

Love the latest string of releases and the moves that the Langchain team has been making. Langchain's React implementation was the the first ever agent i built though their framework couldn't compete

Most AI tools only do one part of building a product. One researches. One designs. One writes code. One generates images. One deploys. You still have to connect everything yourself. II-Agent is differ

Next Tuesday, August 11, we're hosting the August LA Agentic AI Meetup, 5 to 7pm at Gulp in Playa Vista. May drew 70 people. July brought 80. Let's break 100 in August. Come hang out with builders, en

None of this was us. Day 1 (and the first 48 hours) belonged to the open-source community: Generate with it @ComfyUI — native support + official quantized builds, Day 0 Diffusers — the reference Pytho

'OCR is just a feature now. Frontier models will eat it.' We hear this constantly. The data says otherwise. Across three GPT generations, parsing accuracy gained ~24 points, while cost per page 4x'd.