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…
Yesterday
- Datasetcemig-ceia/biblioteca_aneel
Biblioteca ANEEL — metadados e relações 143.762 registros e 16 colunas, preservados integralmente do JSON principal curado. Contém 143.609 entradas de atos e 153 registro
- DatasetLittleFireDragon/S2-Water-Quality
Sentinel-2 Water-Quality Products Daily water-quality products over Lake Pontchartrain, Louisiana, derived from Sentinel-2 MSI L1C imagery. Scenes are atmospherically cor
- DatasetJonesLin/openmmreasoner-sft-concise-cot-314k
OpenMMReasoner-SFT concise CoT (314K) Concise, human-readable rewrites of the reasoning traces in OpenMMReasoner/OpenMMReasoner-SFT-874K: one trace per question, 314,471
- Dataset
- Datasetwebshart/ScaleEdit-12M-metadata
ScaleEdit-12M: metadata-only Webshart companion This repository contains 11,845,023 metadata records across 251 Parquet shards covering every row of InternVL-U/ScaleEdit-
- Datasetlidavidsh/robosmith-assets
RoboSmith assets Build files of the RoboSmith asset library at <id>/<build_hash>/. Each asset keeps the license below. asset license attribution apple_02 Apache-2.0 block
- Dataset
- DatasetFelix-Zhenghao/libero_plus_semantic_action
LIBERO-Plus semantic actions All 14,347 episodes / 2,238,036 action rows are annotated, covering 40 tasks. The dataset keeps its LeRobot v3 layout, original actions, stat
- Dataset
- DatasetMatouK98/math_rlvr_filtered_audited
Math RLVR: audited training and excluded records This is an audited partition of the 35,660 records in clyu/math_rlvr_filtered, configuration rl_final, split train, pinne
- Dataset
- Dataset
- Dataset
- Datasettonystark202609/Annoy-PyEdu-Rs-Raw
Annoy: This should be a paper Title 📑 Paper | 🌐 Project Page | 💾 Released Resources | 📦
- Datasettonystark202609/Annoy-PyEdu-Rs
Annoy: This should be a paper Title 📑 Paper | 🌐 Project Page | 💾 Released Resources | 📦
- Dataset
- Dataset
- Datasetdigas99/so101_tray_train_v2box
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Dataset
- Dataset
- Dataset
- DatasetMahmudisnan/Tulak-tulak
Description Tulak-tulak is a traditional writing system associated with the Mandailing people of North Sumatra, Indonesia. Mandailing script is considered a distinct writ
- Datasetnaver-ai/DroneCATS-Pilot-54K
DroneCATS-Pilot-54K The depth and drone-action data used to fine-tune DroneCATS-Cosmos3-Edge-2B: 53,976 single-turn question-answer rows over frames from NYU Depth v2, SU
- DatasetJackHsieh/4B-general-paragraph-nothink-prefill.k-8.peS2o-humanities-iid.qwen3-ids
4B-general-paragraph-nothink-prefill.k-8.peS2o-humanities-iid.qwen3-ids Thoughts for next-token prediction on k=8 token chunks of JackHsieh/peS2o-humanities-160M-20M-iid
- Datasetvictim434/hfxcol17684
col name sink probe
- Dataset
- Dataset
- Dataset
- Dataset
- Dataset
- DatasetSaagara-Manju-Baiju/clean_so101_ACT_20261023_061301
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Dataset
- Dataset
- Dataset
- Datasetpeichichen/rollout_lab3-eval-test_20261022_205404
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Datasetrainbowrobotics/JIMTOF_newrobot_20261023_000927
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Dataset
- Datasetvictim434/s973-q1
s973q sha-binding probe
- Dataset
- Datasetisaac-ronald-ward/mbari-tank-ood-controlled
mbari-tank-ood-controlled Recorded trajectories (no simulator) packaged as LeRobot splits for world-model training. Generated with quickdraw (data.processors). Observatio
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.