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…
Saturday 24 October
- Dataset
- Dataset
- Dataset
- Datasetchartact-benchmark/ChartAct
ChartAct ChartAct is an interactive benchmark for dynamic chart understanding. It evaluates whether multimodal models and GUI agents can answer chart questions by interac
- DatasetPanda512/record_stackcube_1024
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- DatasetAD12SACZXQW/Annoy-PyEdu-Rs-Raw
Annoy: This should be a paper Title 📑 Paper | 🌐 Project Page | 💾 Released Resources | 📦
- DatasetAD12SACZXQW/Annoy-PyEdu-Rs
Annoy: This should be a paper Title 📑 Paper | 🌐 Project Page | 💾 Released Resources | 📦
- Datasetavishkararjan/llm-reward-gen-quad-dpo-acceptance
llm-reward-gen-quad-dpo-acceptance DPO preference dataset for fine-tuning Qwen2.5-Coder-7B-Instruct to generate quadcopter reward functions with a high validator acceptan
- Datasetmaxylium/anon-reasoning-v5
anon_reasoning_v5 anon_reasoning_v5 is a curated, high-information-density SFT dataset designed to fine-tune language models with an authentic imageboard / anon conversat
- Datasetjonastbrg/mosaic-bench
MOSAIC 199 compositional attack chains across 10 real-world web applications, used to benchmark whether AI coding agents will compose individually-routine tickets into a
- Datasetgavinlaw/setground-kilogram-tasks
SetGround KiloGram元数据 只包含KiloGram自然whole/part标注的无身份字段投影、来源校验和划分盘点；不包含图像、参与者ID、时间或操作轨迹。不包含本项目训练模型或交互收益结果。 来源与许可 上游 https://github.com/lil-lab/kilogram/tree/f2cd80201eedc5f
- DatasetImk1dding/phishing-email-dataset.csv
Phishing Email Dataset This dataset on Hugging Face is a direct copy of the 'Phishing Email Detection' dataset from Kaggle, shared under the GNU Lesser General Public Lic
- Datasetjyohukuchan/qwen38-flash-next-bf16-logits
Qwen3.8-Flash-Next BF16 reference logits (KL evaluation set) Full-vocabulary FP32 logits of the official BF16 model Qwen/Qwen3.8-Flash-Next (revision de4b8e4d43b917e77067
- Datasetlandauros/output1
output1 Archive of the full deployment_traces directory. Upload is in progress until UPLOAD_COMPLETE.json exists. Source files: 277,107 Source size: 665,043,670,941 bytes
- Datasetopen-athena/Marin-Snowball-GRPO-Results
Marin Snowball GRPO results Two 20-update teacher-repair GRPO pilots completed successfully but did not improve observed held-out identity full-success counts. Scores bel
- Datasetopen-athena/Marin-Snowball-GRPO-Identity
Marin Snowball GRPO identity-repair data Research data from two unsuccessful 20-update identity-transfer pilots, October 24, 2026. Public with automatic gated access, con
- Dataset
- Dataset
- DatasetMralisanreynz1200/opus5-5-doctor-patient-conversations-all-human-diseases
Opus-5.5 generated Doctor-Patient Conversations for All Human Diseases The sequel to nisten/opus-doctor-patient-conversations-all-human-diseases (Opus 4.8). Same disease
- Datasetteamsleeping/creative
Introducing Creative, a human-written literary dataset developed by Sleeping AI, designed for fine-tuning language models towards stylistic adaptation, narrative comprehe
- Datasetcoral-bricks-ai/coralbricks-use-case-traces
CoralBricks use-case traces The full agent traces behind the CoralBricks use cases: Claude Code running on DeepSeek V4.1 Flash, served by CoralBricks, doing real jobs end
- Datasetzjumty/so101-act-side-wrist-joint103
SO-101 side+wrist pick-and-place demonstrations (joint103) These are 103 demonstrations of the SO-101 arm picking an orange cup and placing it on a fixed target in MuJoCo
- DatasetYuChuang1205/VersaCoT-10K-RAD-Version
VersaCoT-10K-RAD is the rationale-augmented version of VersaCoT-10K, developed for multimodal Chain-of-Thought (MCoT) reasoning and reasoning-state refinement. It builds
- DatasetPEN2044/setD435_aser_raznoobAAAA_OHIBKI_20261024_171802
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "shape":
- Dataset
- Datasetaqxlogistics/united-kingdom-and-australia-audiophile-analog-synthesizers-air-cargo-2026
United Kingdom & Australia Audiophile Analog Synthesizers Air Cargo (2026 Benchmark) Authoritative dataset benchmarking transatlantic and transpacific climate-controlled
- Datasetaqxlogistics/bahrain-and-gcc-medical-ultrasound-transducers-and-diagnostic-probes-freight-2026
Bahrain & GCC Medical Ultrasound Transducers & Diagnostic Probes Freight (2026 Benchmark) Authoritative dataset benchmarking delicate acoustic piezoelectric crystal freig
- Datasetaqxlogistics/kuwait-and-qatar-marine-yacht-hydraulic-thrusters-and-stabilizers-freight-2026
Kuwait & Qatar Marine Yacht Hydraulic Thrusters & Stabilizers Freight (2026 Benchmark) Authoritative dataset benchmarking maritime heavy propulsion logistics, hydraulic b
- Dataset
- Datasetaqxlogistics/saudi-and-uae-commercial-3d-resin-and-sls-printers-freight-2026
Saudi Arabia & UAE Commercial 3D Resin & SLS Printers Freight (2026 Benchmark) Authoritative dataset benchmarking international air cargo tariffs, delicate galvanometer l
- Datasetaqxlogistics/saudi-arabia-and-gcc-4x4-suspension-and-remote-reservoir-shocks-freight-2026
Saudi Arabia & GCC 4x4 Remote Reservoir Shocks & Suspension Freight (2026 Benchmark) Authoritative dataset benchmarking international air freight tariffs, nitrogen-charge
- Dataset
- Dataset
- Dataset
- Dataset
- Dataset
- Dataset
- Dataset
- Dataset
- Datasetanyantudre/moore-tts-samples
Mooré TTS samples Listening samples for the Mooré text-to-speech models in our benchmark. The model cards link here so anyone can listen without requesting access to the
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.