Search: “jeremy-howard--x”

Degrading performance on ML research *without telling the user* is shockingly hostile and a terrible look. That could silently damage all sorts of work, including some of my own. Also the type of thin

GPU MODE has powered much of the public GPU kernel work online, with a permissive license from day one and generous credit from researchers, NVIDIA, AMD, and others. Today we’re moving our datasets to

I don't really want to have to go to bat against Anthropic, but they've just been unnecessarily antagonistic to all of China, then not so subtly to open weight models, and now more broadly open AI res

This post critiques opaque AI moderation systems that penalize users without clear explanation, drawing parallels to how some companies might arbitrarily restrict customer access based on undisclosed

It's very cool that Apple shipped a 20B parameter on-device. You can't put 20B parameters in RAM at any reasonable precision. To make it work they are using pretty exotic architecture by today's stand

June 9th Researcher Reciprocity License 'if you train on it, you let us generate - reverse terms of use void' Status quo 1. We teach frontier devs with ICLR/NeurIPS papers, OSS Github contributions 2.

mythos will be bad ON PURPOSE on ai 'frontier llm research' tasks, this is very very sad for the research community also the fact that this is un purpose not visible to the user is crazy Introducing C

One of my personal favorite features announced at WWDC will I suspect be a sleeper hit: container machines, allowing your Mac to run a lightweight, persistent Linux environment with your home director

Open Source is the only way... First they came for the model builders... I feel we're getting a glimpse of a future where AI is only provided to a privileged few, and that's not a future I want to liv

Jeremy Howard argues that calls to pause AI research are selectively applied, with restrictions primarily targeting independent researchers while well-resourced labs continue development, creating an

Also, those who focus on using AI to help improve the skills of themselves and their teams will be diamonds in great demand, since they will be the rare A++ players in a sea of mediocrity. There will

Does a token buy you more or less now than it did a few months ago? We built a consumer price index (CPI) for AI coding output from Anthropic's Opus 4.6 model in SWE-chat, Feb 5–Apr 15, 2026. What we

This is super big I think this is the first useful speculative decoding method deployed on a big quasi frontier model Massive unlock @fi56622380 🚀 1,000+ TOKENS/S ON A 1T MODEL! 🚀 We are thrilled to r

These were some magical results from distillation by @geoffreyhinton that really shocked me when I first saw them, and TBH I still don’t fully understand it even to this date https://www.ttic.edu/dl/d

American AI startups have increasingly shifted toward adopting Chinese AI models since the beginning of the year, representing a notable change in technology sourcing preferences. This trend likely re

Jeremy Howard discusses a significant increase in output capabilities driven by agentic AI systems, while noting that adoption rates remain completely flat across the board. The observation suggests a

This post likely discusses Jeremy Howard's commentary on software engineering efficiency and the goal of shipping significantly reduced code volumes—possibly advocating for more concise, optimized cod

A few months ago, I found an anonymous sockpuppet account linked to the OpenAI/a16z super PAC. Now, @TaylorLorenz and I have uncovered two more — and they're even more brazen than the first. https://x

Jeremy Howard announced the release of Vui, an open-source voice mode text-to-speech (TTS) model with 300 million parameters that can run on consumer GPUs and Apple Silicon. The model features context

Climbing with no distillation, like the Big Boys do, has been super fun! Read the tech report for a taste of our ̶s̶u̶f̶f̶e̶r̶i̶n̶g̶ journey Super excited to announce seven new world-class MAI models

Just saving this here to document a story and as a self reflection on whether AI is really making me more productive Yesterday morning I found a way to complete the new HVM approach, that is much fast

Nemotron 3 Ultra is NVIDIA's latest language model featuring significant improvements in speed (5X faster) and cost efficiency (30% cheaper) compared to previous versions, positioning it as a frontier

You can work 5 days a week and succeed as a startup. Mercury has done that from day 0 and we are valued @ $5.2bn 7 years after launch. I have been an entrepreneur for 20 years and raised 3 kids while

what a wonderful project: parakeet.cpp https://github.com/mudler/parakeet.cpp GGML based parakeet inference pipeline that's 2x faster than my ONNX parakeet pipeline on Apple Silicon! (Needed a few loc

Has @AnthropicAI completely given up on making API usage reasonably-priced? Following the token-usage changes recently they announced various updates to *subscription* usage to make it more reasonable

Jeremy Howard expresses positive views on GPT-5.5, stating he has recently switched to using it as his primary model and finds it nearly comparable to Anthropic's Opus 4.6 and 4.7 while offering signi

My MLSys keynote on AI writing systems code got more interest than I expected. The recording will take a while, so in the finest tradition of AI labs sharing blog posts, we’re starting the Core Automa

Worked on some code this morning using Opus 4.8 and so far I'm really liking it. Much more cooperative than 4.7 and less 'over agentic'. Stops and asks for my input when needed in places 4.7 (and GPT

65B private round More than double the size of the largest IPO ever We've raised 65 billion in Series H funding at a $965 billion post-money valuation, led by @AltimeterCap, Dragoneer, @Greenoaks, and

Fascinating results + Anthropic running away with it right now + So many people want to start their own company + Google over OpenAI + Vercel, Linear, Every, PostHog overperforming A great list if you

Fun fact: Australia is basically Scandinavia with the Sahara Desert bolted on. I LOVE this visualisation. Everyone imagines nature and the outback, but they don't realise just how urbanised we are. Cr

glad to know Mythos' safety concerns have been addressed right as Anthropic also secured tens of billions in inference compute 👍 JUST IN: Anthropic announces it will roll out Claude Mythos “in the com

wait… if most people think 5.5 is better than 4.7, i assume that’s due to terminal coding benchmark… 4.8 is still outperformed by 5.5 Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper ju

Behind the MiMo API Price Reduction: The deepest price cut, up to 99%, is for Input (Cache Hit). The core reason is our inference framework now supports hierarchical KV cache optimization for SWA. Pro

Jeremy Howard comments on the Xiaomi MiMo v2.5 model, highlighting its exceptional value proposition and cost-effectiveness for prompt processing. The post appears to include comparative pricing data

Jeremy Howard comments that large language models released 6 months ago feel comparable in capability to recently released models, suggesting that the pace of improvement in model development may be s

recommended reading. i too am very done with people anthropomorphizing a bunch of matrices on a GPU cluster, especially if the same people do not give two fucks about actual human beings. More musings

Nothing disappoints me more than people saying we should stop progress because peoples meaning depends on that monotonous labor, as if humanitys highest purpose is filling Excel sheets or stocking she

RWKV-7 G1g is here: the world's best pure RNN LLM, and a competitive LLM in general. Try https://huggingface.co/spaces/BlinkDL/RWKV-Gradio-2 for bsz16 7B inference. G1h in June 🙂 p.s. const 15000+tps

this has been my experience as well. there definitely were improvements, specifically wrt shell based computer use, but also regressions, especially in the last 3 version bumps of flicker and gerperte

Gated DeltaNet-2 is almost exactly RWKV-7's DPLR recurrence, not acknowledging the elephant in the room 🙂 Gated DeltaNet-2 is here. 🚀 🔥 New paper: Gated DeltaNet-2: Decoupling Erase and Write in Linea

Gemini Flash 3.5 is such a disappointing model. It's intelligence and speed is awesome. Absolutely amazing. But it's been trained to max evals, not to be helpful to humans. It goes off and does random

Jeremy Howard comments on comparative performance trends between GPT 5.5 and Claude models, noting that GPT 5.5 appears to be improving in a particular capability while Claude models are declining in

Jeremy Howard argues that AI model developers at OpenAI and Anthropic should prioritize breaking backwards compatibility during the prototyping phase rather than maintaining it, suggesting that backwa

My concern for the AI era, or at least this phase of it, is that a generation is being taught that 'close enough' is just fine. Take @AnthropicAI for example. Text wrapping in Claude Code has been bro

DeepSeek has announced a permanent discount for its DeepSeek-V4-Pro model, encouraging developers to build and innovate with the platform. The announcement was made via social media and emphasizes the