2026
10,570 articles · 132 days · 10 categories · avg 80 per day
OpenAI Releases GPT-Live-1.1: New Voices in 6 Languages and Fewer "checking..." Pauses
GPT-Live-1.1 is now available to all users, with new voices for French, German, Hindi, Mexican Spanish, Korean and Japanese, plus more accurate accents.
Tadpole Proposes a Foundation Model for Physics: From Prediction to Understanding
Tadpole brings foundation-model logic to science: one model instead of separate networks for weather, airflow, material deformation and electromagnetism.
Stanford Online: AI Can Estimate Impact and Cost, but a Human Sets the Priority
Stanford's Anand Subramani and Laurel CPO Jiaona Zhang reduce roadmap prioritization to three steps: AI helps with two, and the third stays with a human.
SpecTr-GBV at EMNLP 2026 Combines Multi-Draft and Block Verification to Speed Up LLMs
AIPOCH Open-Sources Open-Science, a Lab Notebook for "Vibe Research" with AI
Honeycomb Gives Video Models Constant-Size Memory for Loop Closure
New Survey Maps AI in Humanitarian Work: Six Tasks, From Weather to Refugees
ARC-SORT hits 81.82 HOTA on SportsMOT and runs at 50 fps
The training-free ARC-SORT tracker scores 81.82 HOTA on SportsMOT with the fewest ID switches among 69 entries, and runs at about 50 fps on a laptop RTX 3060.
SciΠvot benchmark: the best LLM picks the right plan 63.5% of the time, versus 73.4% for humans
A benchmark of 248 senior-researcher corrections finds the best model picks the right call 63.5% of the time; fresh researchers hit 73.4%. Project context tends to…
A Frozen 16 MB Vocabulary Matches a Trained One in Small LLMs
NVIDIA RTX Spark N1X Runs a Local Qwen3.8 27B Agent in 18.7 Minutes
Vera Rubin NVL72 beats GB300 by up to 5.7x in tokens per megawatt in AA-AgentPerf
NVIDIA releases Nemotron 3.5 Super: 26 AA points and 630 tokens/s
Kolab: 115 Hours of Recordings Show That AI Agents Don't Understand Their Human Colleagues
Minnesota NLP researchers recorded 50 professionals from 40 occupations working with AI agents. Frontier models overestimate how well that collaboration goes.
DialFork measured procedural hallucinations in agents across 36,278 dialogues
The DialFork benchmark catches agents that sound smooth and factually correct but drift off script. It includes 36,278 dialogues, 24 tasks, 12 domains, and two new…
OpenAI Engineer Hid a "Bug" in Toilet Paper and Got Caught Within 48 Hours
Fireship tells of an OpenAI employee who built a listening device tied to his wife's ChatGPT account, hid it behind a TV in the executive meeting room, and was…
Tsinghua Unveils SRHarness and MechBench: AI Agents That Look for Formulas and Mechanisms
Tsinghua Unveils Jev-LDE: Edit ICL Examples Once, Boost 1-Shot Accuracy
Recursia Beats Nvidia ASPIRE by 10 Points on Long-Horizon Robot Tasks
ETH and Roche Train AI to Predict Where C–H Borylation Will Occur on a Molecule
PoS Method Catches AI Agents Stuck in Loops of False Belief: 36 Repeats in ALFWorld
An ALFWorld agent forgot to heat a cup and checked the cabinet 36 times in a row until its action budget ran out. A new paper calls this Belief Trapping and…
Ego Data Gets a "Full Body": A New Scaling Path for World Models
Robot data collection is shifting from expensive platforms to lightweight formats. Jiqizhixin examines the idea of extending ego recordings with whole-body motion…
SAE-IDF Matches Symbols Across Different Systems and Beats a Baseline by Up to 45 Points
ReaDy speeds up hybrid LLM decoding 3.16x on a single B200
DAGFormer Plans Connections Between Layers and Cuts Perplexity by 17–25%
Researchers Find a 0.4–1.1% Window in Pythia Pretraining Where the Readout Reorganizes
MULTIFINGROUND: LLMs Recognize Accounting Concepts With 99% Accuracy but Miss Facts
A new benchmark on filings from 90 issuers: models identify the concept with up to 99.4% accuracy, yet in 26.6–41.9% of cases they miss at least one related number.
Interlock teaches agents to finish the job: up to +6.64 points on JobBench
The Interlock framework synthesizes long-horizon tasks for LLM agents. Fine-tuning two Qwen3.6 models on their own trajectories gave +4.00–4.50 pp on…
BlenderWorld: GPT-6-Astra Beats Opus-5 in 3D Benchmark with a Score of 79.9
The new BlenderWorld benchmark ran 13 frontier models through 223 3D creation tasks in Blender. GPT-6-Astra leads with 79.9, but spatial coherence still lags.
Agent teams for $155 instead of $510: Cohere finds where delegation actually pays off
WinARDet fixes looping in MLLM detectors: F1 rises from 32.16 to 39.36
GAFU pulls knowledge out of generation in UMMs and lifts VStar-Bench by 20%
