RSI-Anything Cuts Expensive AI Feedback Calls by Predicting Actions First
A new framework lets AI agents improve themselves across code, GUI, lab planning, and chemistry by using predictive world models to filter their own actions.
20 stories
A new framework lets AI agents improve themselves across code, GUI, lab planning, and chemistry by using predictive world models to filter their own actions.
Vals AI pitted GPT 6 Sol and Claude Opus 5.5 against themselves on Vibe Code Bench, running each as a solo coder and as a multi-agent team.
HeyGen's new in-house TTS model debuts on top of the Controlled Voice Arena, beating Qwen-Audio-3.1-TTS-Plus and ElevenLabs' Eleven v4 Turbo.
xAI's Grok Bot agent now ships with a native email inbox so it can sign up for services, verify accounts, and run outreach without borrowing yours.
A 30-edit real estate staging stress test shows how much of the frame each top image model actually touches, and how fast errors compound.
Claude's dynamic workflows exit research preview with a new multiagent API type, letting a lead agent plan and orchestrate up to 1,000 subagents per run.
Factory folds in YC-backed Mentlio to bring intelligent model routing and token-level ROI measurement into its autonomous software engineering platform.
Jina ships a 1.74B multimodal embedding model that encodes text, images, video, and audio into one shared vector space without reindexing.
Dao AI Lab and NVIDIA show that caching SSM inputs instead of recurrent state cuts memory traffic and unlocks quantization for hybrid models.
A 4B decision-native model from NUS handles 2,048 candidates, zero decision flips under reordering, and 20x speedups via cross-request caching.
Ai2 replaced its priority-based GPU scheduler with time budgets and fair-share allocation, cutting median queue wait from 5 minutes to 24 seconds.
A new benchmark of 50 unpublished research problems finds frontier agents solving nearly every one for under three dollars each, with little help from the web.
Cactus Compute's Whistle squeezes multilingual speech-to-text into a 16.9 MB CPU file that reportedly beats Whisper base with 9x less size and 6x speed.
A set of Claude Code and Codex skills that automate shot-by-shot film analysis, using ffmpeg for hard numbers and the model only for judgment calls.
A 3-bit EXL3 quant squeezes a 753B uncensored GLM-5.3 Mixture of Experts into 273 GiB, making local inference possible on four RTX PRO 6000s.
A 400MB anime-style LoRA for Qwen Image 2.1 brings curated aesthetics, character sheet conditioning, and open weights without a gatekeeper.
Tencent's 5B omni-modal parser handles documents, charts, geometry, audio and video into one structured JSON schema, rivaling Gemini-3-Pro.
Cloudflare adds a Qwen3-Omni mixture-of-experts variant that scores audio and video alongside text and images in one pass, with no token generation.
NVIDIA open sourced its NURBS boundary-representation CAD kernel under Apache 2.0, giving developers an exact-geometry solid modeling library that reads and writes OpenUSD natively.
A new study dissects why continuous diffusion text models work, replacing self-conditioning with cheaper tricks that push TinyGSM accuracy from 4% to 50%.