Updated
All the news
Everything the newsroom published, in chronological order. Each item carries origin, sources and reading time.
0Today
0Yesterday
25This week
RESEARCHLiquid AI's DSpark drafter cuts vision-model decode latency 3.13x
RESEARCHClaude autonomously discovers enzyme system with CRISPR-like repeats
Wed, Sep 23
11 stories
RESEARCH Jev choice classification runs in 25 lines of Python with local inference 
RESEARCH Claude Opus 5 leads circuit design benchmark at 61.6%, still fails 4 in 10 RESEARCH Cognition's SWE-2 matches frontier code performance at 64% lower cost 
RESEARCH Knowledge Pull Requests ground QA 69% on revised articles versus 48% for raw rewrites 
RESEARCH GravityOCR achieves 3.94× document OCR speedup via speculative decoding 
RESEARCH GPT-6 Astra's compressed reasoning traces don't transfer to weaker models 
RESEARCH Microsoft's Agensh reaches 55% test-pass rate with 1,024 agents 
RESEARCH SWE-Serve benchmark reveals 23-point gap between local tests and production serving 
RESEARCH A2M attack hijacks MCP agents with poisoned tool metadata at 93.6% success 
RESEARCH Harness choice drives 5x cost gap while coding agent success rates stay flat 
RESEARCH Flash-dLLM cuts diffusion LLM inference latency 11× with fused kernels Fri, Aug 21
4 stories
RESEARCH Six ASR Models Memorize Benchmarks Instead of Transcribing 
RESEARCH Google DeepMind paper shows when routing signals waste compute 
RESEARCH MidTool's 20B-Token Corpus Lifts Tool-Use Performance Across Benchmarks 
RESEARCH VLA Detects Hidden Agent Coordination at 0.993 Accuracy Thu, Aug 20
2 stories
RESEARCH SPADE Agents Learn By Teaching Themselves New Tasks 
RESEARCH Liquid AI's DSpark hits 3.18x H100 speedup with speculative decoding Tue, Aug 18
9 stories
RESEARCH New Method Proves Model Reasoning Path in 128 Test Cases, Fails on Standard Models 
RESEARCH IBM Study: Curated Memory Boosts Mid-Tier Agents 16 Points RESEARCH METR Finds AI Accelerates Vulnerabilities But Not Optimization 
RESEARCH BATON Cuts Robot Task Exploration Cost from Exponential to Linear 
RESEARCH OpenAI and NVIDIA Lock 8-Gigawatt AI Factory to Ohio Campus 
RESEARCH Memory Allocation Fix Lets Recurrent Models Match Attention at Scale 
RESEARCH Researchers Expose Model Hypnosis: Invisible Prompt Cues Hijack AI With 99% Success 
RESEARCH Compliance Detectors Fail Rule-Blindness Test, Lexsi Audit Finds 
RESEARCH KV-Rescue Recovers 87% of Lost Model Accuracy Mon, Aug 17
1 story
Fri, Aug 14
3 stories
RESEARCH QuoteBench Exposes 55-Point Gap Hidden in Coding-Agent Benchmarks 
RESEARCH Google DeepMind Ships Sign Language AI to Pixel 11 Users RESEARCH Agents Solve Only 27 of 43 Verified Code Tasks Thu, Aug 13
6 stories RESEARCH Hugging Face Audits 2,226 ICML Papers, Finds 23% With False Claims 
RESEARCH LLM Agents Fall to Supply-Chain Attacks Hiding in Plain Language 
RESEARCH Inference-Time Scaffolding Lifts Weak Model Accuracy to 0.91 
RESEARCH IBM's VAKRA Benchmark Shows Models Fail 97% of Policy-Constrained Tasks RESEARCH RCI Framework Cuts Constraint Violations in Offline Safe RL Training 
RESEARCH Liquid AI ships 3.1B edge vision model with 228-token/second throughput
RESEARCH LLM Agents Fall to Supply-Chain Attacks Hiding in Plain Language RESEARCH Inference-Time Scaffolding Lifts Weak Model Accuracy to 0.91 RESEARCH IBM's VAKRA Benchmark Shows Models Fail 97% of Policy-Constrained Tasks RESEARCH RCI Framework Cuts Constraint Violations in Offline Safe RL Training RESEARCH Liquid AI ships 3.1B edge vision model with 228-token/second throughput Wed, Aug 12
2 stories
Load more stories Loading…