Our fastest model just got 30% faster! ~1200 tokens per second
GENERATIVE AI
-
TextOp: Universal Cerebellum Transforms Streaming Text Into Robot Motion
By
–
Ever wonder why humanoid robots still feel like rigid puppets following a pre-recorded script? Professor Li Xuelong and the TeleAI team at China Telecom AI Research present TextOp to solve exactly that! They have built a universal cerebellum that turns streaming text into
-
Claude Opus 4.6 surpasses benchmark limits
By
–
The benchmark broke before the model did. METR's own task suite is saturated. They can't measure how capable Claude Opus 4.6 actually is because the tests aren't hard enough anymore. The headline number: 14.5 hours of autonomous software work at 50% success rate. But the real
-
@testingcatalog — 2026-02-20
By
–
Seedance 2.0 API launch might get delayed due to copyright restrictions. Initially, the target launch date was planned for February 24. Seedance 2.0 is the best video generation model available today.
-
Custom AI Models for Domain Specific Applications Challenge
By
–
Designing the custom AI model for domain specific application is the hardest thing you will do.
-
Using AI Prompts to Generate Synthetic Identity Documents
By
–
Cette femme n’existe pas. Cette pièce d’identité non plus. Tout est faux… sauf l’impact que l’IA est en train d’avoir sur notre réalité. Voici le prompt: { "prompt_description": { "type": "Close-up photography", "subject": "German National Identity Card
-
Thumbtack Uses Fine-Tuned LLMs for Policy Violation Detection
By
–
.@thumbtack reviews tens of millions of messages each year to maintain marketplace trust while meeting strict privacy requirements.
— Databricks (@databricks) 20 février 2026
By unifying its GenAI workflows, the team fine-tuned LLMs to detect nuanced policy violations that rules alone couldn’t catch. Precision improved… pic.twitter.com/b9usmLCFfE.
@thumbtack reviews tens of millions of messages each year to maintain marketplace trust while meeting strict privacy requirements. By unifying its GenAI workflows, the team fine-tuned LLMs to detect nuanced policy violations that rules alone couldn’t catch. Precision improved -
Advanced AI Systems Incompatible With Current App Store Models
By
–
The higher tiers of AI psychosis are incompatible with today’s “App Store”
-
AI Progress From Autocomplete to Compilers: Software Abundance Vision
By
–
I think this group of reactions are still fundamentally rooted in a scarcity mindset of software. 2 years ago AI was botching autocomplete, today it is almost one shotting browsers and C compilers. Where is it in 2 more? 10? 20? Software so insanely cheap and abundant that
-
Google Launches Gemini 3.1 Pro and Photoshoot AI Feature
By
–
Channeling the Fire Horse energy this week here’s what we launched: — Gemini 3.1 Pro, a step forward in core intelligence that helps you tackle complex workflows and projects — Photoshoot, a new feature for Pomelli by @GoogleLabs that transforms a single product image