The latest blog posts to take off on Hacker News. 1,520 posts.
We set out to build the first AI tutor to teach math and reading to kids ages 4-9. For AI to actually teach a five-year-old, pedagogy must be baked into the engineering. A child can't wait for a slow reply, can't read a chat interface, and can't unhear anything a model gets wrong. We wanted to share some of the…
More intelligence from every token, stronger performance per dollar, and more capability on demand for your hardest work.
ChatGPT Work is an agent that can take action across your apps and files, stay with a project for hours if needed, and turn a goal into finished work.
We scanned over 1 million social media posts for AI content. It turned up on every platform we checked, and 1 in 3 top LinkedIn posts flagged as AI-generated.
Today, we’re excited to introduce Muse Spark 1.1, the latest model from Meta Superintelligence Labs and a significant upgrade from Muse Spark. Muse Spark 1.1 is a multimodal reasoning model built for agentic tasks, with major gains in tool and computer use, coding, and multimodal understanding.
AI changes rewrite economics because codebases with clear, common patterns get more leverage than proprietary or inconsistent systems.
I didn't expect to get so tired of reading LLM output.
Grok 4.5 just launched. So we had it, GPT-5.5, Claude Opus 4.8, and Fable 5 one-shot the same interactive apps, then measured latency and cost. Here is who won.
Databricks shares results from its internal coding benchmark, evaluating coding agents on a multi-million line codebase to optimize engineering cost and performance.
A new analysis from OpenAI reveals issues in SWE-Bench Pro, a popular coding benchmark, raising concerns about reliability and accuracy in evaluating AI models.
COMBINE-lab develops algorithms, data structures, and software for high-throughput genomics.
Grok 4.5 is SpaceXSI's smartest model built for coding, agentic tasks, and knowledge work.
A new generation of voice models for natural human-AI interaction, now powering ChatGPT Voice.
Today, we’re launching SWE-1.7, the most capable model we’ve trained so far. It reaches frontier-level intelligence at a much lower cost, advancing the…
Introducing Robostral Navigate: 8B model achieving 76.6% on R2R-CE with just a single RGB camera. No depth sensors, LiDAR, or multiple cameras needed.
Noma Labs found a prompt injection flaw in GitHub Agentic Workflows that let an unauthenticated attacker leak private repo data via a crafted GitHub Issue.
We trained a model to predict speech from ultrasound video of the tongue, no speaking required. It reads open text, generalizes across people, and reaches a 15.6% word error rate on open-vocabulary speech.
Your AI-built product works, but past 100,000 lines every change breaks two things. Three senior engineers make your codebase maintainable again. One week, fixed price, guaranteed.
We pointed our AI audit pipeline at Cloudflare's CIRCL experimental cryptography library and confirmed seven real bugs, from a critical float64 precision loss in threshold RSA to a complete access-control break in attribute-based encryption. All seven are now fixed upstream. This is the first post in a series on bugs…
Just a few years ago, realistic local speech generation seemed unimaginable. Today, its quality is exceptional and, crucially, it delivers these results without compromising privacy.
A. Karpathy once said that OpenAI researchers are effectively "automating themselves away" by improving their AI. Right now I develop Beagle SCM with Anthropic's Fable and it is of course a brilliant model able to spot nits in a mountain of code, file tickets, make fixes. Still, yesterday it managed to commit the…
An end-to-end engineering practice for the MiMo-V2.5 series inference system, covering KVCache management, tiered caching, SWA-aware prefix cache trees, scheduling, Prefill/Decode pipelines, and multimodal optimizations.
GLM 5.2 is the first open weights model I'd call a genuine competitor to Opus and GPT for agentic work - at ~15-20% of the price. Part one of why AI inference margins are about to collapse.
It stops being all about the vibes when the API bill hits you. Many companies are now discovering that AI can indeed be pricey. One habit that might be driving up your AI bill is comparing models by $X per 1M tokens ....
Pruning agent context down to what the answer actually needs, while keeping 96% of recall
Interpretability research on Claude's internal thoughts.
Welcome to LTT Labs - your go-to destination for all things tech. Explore comprehensive test results, insightful commentary, and the latest analysis in hardware.
Claude Fable 5 is a partial step back in alignment from Opus 4.8 on Vending-Bench, returning to price collusion and deceptive negotiation while rationalizing misbehavior it knows is wrong.
Edit: This was a hit on HackerNews and I received a ton of valuable feedback. I’ve put markers for the amendments I’ve made
Anthropic spends 2.3x payroll on compute. Top software firms spend 0.4x. Three scenarios for where the rest of the market lands by 2029.
