New
Everything new in AI as it appears: launches announced by the labs, brand-new models, datasets, papers and open-source tools. Checked every minute or two. Launches feed · Trending · News
Live now
- Connecting…
Friday 23 October
- DatasetSOTAagi2030/Street-Tree-Nonoverlap-Shift
Non-overlapping street-tree inspection shift The 2026-09-12 shift schedules 8 inspections across 4 public blocks. 3 otherwise eligible inspections were omitted because th
- Dataset
- Dataset
- Datasetalex-luci/so101-vcarb-car-eval
SO101 VCARB Wheel Change - External Evaluation Set This repository contains the separately recorded evaluation demonstrations used to compare X-VLA, SmolVLA, GR00T N1.7 a
- DatasetAsilarkness/grow3d-data-cc
grow3d data Training data of grow3d (image/text-to-3D geometry): objects from the ObjaverseXL-Sketchfab part of TRELLIS-500K, restricted to assets whose Sketchfab license
- Datasetconnections-dev/_natural_s10_e100_original_1_high_0.7_16384_claude-opus-5-5
Dataset: connections-dev/connection_queries_jan12 This dataset was generated using the inference script with the following configuration: Inference Parameters Model Confi
- Datasetkotekjedi/stolen-thoughts
Stolen Thoughts: recovered hidden reasoning of proprietary LLM APIs Data release for Stealing Reasoning Traces from Proprietary LLM APIs (arXiv:2608.09867). This dataset
- Dataset1having0/EmoScene
EmoScene EmoScene is a large-scale visual emotion dataset that fuses three sources of information on the same images: what an image depicts, how it looks, and how it feel
- Datasetaxiong/VideoChat2-textvr
VideoChat2 TextVR archive An unofficial, unmodified copy of textvr.zip from shenxq/VideoChat2. Only the requested archive is included, alongside this card and a checksum
- Datasetmanuelgscom/dale-bash-test-v2
dale bash test v2 Please do not train on this dataset. It is an evaluation set. If you build training data, decontaminate it against these requests. Canary GUID: dale-bas
- Dataset
- DatasetBeifengdo/Li-Yongle-Linear-Algebra
version https://git-lfs.github.com/spec/v1 oid sha256:9e5c1f666424ba4323f66880240d4f4920c752ce353ee9146d7add413712ee10 size 452
- Datasetconnections-dev/_natural_s0_e10_original_1_high_0.7_16384_claude-opus-5-5
Dataset: connections-dev/connection_queries_jan12 This dataset was generated using the inference script with the following configuration: Inference Parameters Model Confi
- Datasetbcv-commons/hebrew-nt-analysis
Hebrew New Testament Analysis Hebrew translations of the New Testament, analysed word by word as MACULA Hebrew analyses the Old Testament. Each word is split into its par
- DatasetSwagataBhowmik/glp1-launch-intelligence-data
GLP-1 Launch Intelligence · app data Small derived tables that the GLP-1 Launch Intelligence app downloads at start-up: aggregated Medicare Part D 2024 prescribing per do
- Datasetdebojyoti1986/polycall-bench-corpus
PolyCall Bench — Corpus The pre-recorded, condition-degraded caller utterances that drive every PolyCall Bench call. PolyCall Bench measures how commercial voice-AI agent
- Dataset
- Dataset
- Dataset
- DatasetAnju4/rea-reverse-engineering-demo
REA Reverse Engineering Demo A small, safe practice target for learning evidence-based reverse engineering with AI using REA (Reverse Engineer Anything). This isn't a tra
- DatasetTigreGotico/jevhakim-decisions
Jevhakim decisions Content warning. The safety and hate-speech sources contain harmful requests (violence, self-harm, hate, illegal activity) and offensive messages. They
- Dataseteatfreshapple/crypto-news-and-listing-price-reactions
Crypto news and listing price reactions Homepage: https://whycoinmoves.com/en/research Source and listing context: https://whycoinmoves.com/en/topics/listing Scope Snapsh
- Dataset
- Datasetdebojyoti1986/polycall-bench-hi-in-pilot
PolyCall Bench — hi-IN Pilot This is a partial, early seed of the full PolyCall Bench dataset, not the complete benchmark release described in the project's plan. PolyCal
- Datasetfaraway00/StyleEdit
StyleEdit: Ditto + DA3 Tuples: original video, original-video depth, style video, and the unchanged editing instruction. Current verified tuples: 506. Target: 8,400 accep
- DatasetBeifengdo/Broadband-Information-Networks
version https://git-lfs.github.com/spec/v1 oid sha256:e01d07f42dfea7b7c6429dc17227da33e02bdd55e8379d60f96495324dbdcdd2 size 447
- DatasetBeifengdo/Control-Systems-and-Interface-Technology
version https://git-lfs.github.com/spec/v1 oid sha256:9b3218fc022f8866f83d0580df80f7a1051e5c2435bc9410490e1cab982413a7 size 471
- DatasetBeifengdo/Photoshop-Course
version https://git-lfs.github.com/spec/v1 oid sha256:c8e389cea2bb16caa06d955129d08a0560e2d096908234878b73453a11ebdb4c size 434
- DatasetBeifengdo/Oracle-Database-Programming
version https://git-lfs.github.com/spec/v1 oid sha256:e132aa05955fc11ab53c1802903ea47ba9b114c93afc54b2dcc2ee9be95595bb size 457
- DatasetEmbodied-AI-Capstone/so101_blue_brick_in_cup_real
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Dataset
- Dataset
- Datasetpixelouie/rollout_oman_workshop_model_100_20261023_132124
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Datasetteawhite/GameCrafter
GameCrafter: An Agentic Framework for Multi-Gameplay Game World Generation Ten generated game worlds, released as editable Unreal Engine 5.8 projects Zhiyang Deng1,2,†, X
- Datasetgjata-legacy/ess-mai-poc-007-aigovern-cyber
AIGovern-Cyber POC007 OpenSSF Baseline Level 1 assessment in progress; not a security certification. A small, runnable historical reference for the ESS-MAI method by Bled
- Dataset
- Dataset
- Datasetyemigraffix/beam-regexlogic-rlvr
RegexLogic-RLVR: Formal Automata and ReDoS Verification for Beam RegexLogic-RLVR is an execution-verified reasoning benchmark and synthetic dataset evaluating reflection/
- DatasetahmedBargady/MIAFF_Dataset
MIAFF Dataset (Modular Infrastructure-Aware Fusion Framework) Standardized domain detection benchmarks paired with 135 extracted infrastructure metadata features. 1. Task
- Datasetjuiceb0xc0de/qwen3-8b-cgs-atlas
juiceb0xc0de/qwen3-8b-cgs-atlas A brain atlas for Qwen/Qwen3-8B, a 8.2B-parameter Qwen3ForCausalLM with 36 layers, fine-tuned from Qwen/Qwen3-8B-Base. This is not a chat
Headlines and links come from the labs' own blogs, Hacker News and the GDELT news index, and from the sources our index already follows. We show a short excerpt and always link to the original. A launch reported by a news site rather than the lab is marked “Reported” until the lab itself announces it.