Resumen diario de blogs de proveedores, artículos seleccionados, modelos en tendencia e incidentes de servicio.
Los artículos, resúmenes, fuentes y elementos del archivo se muestran tal como se publicaron, en inglés.
STGformer: Efficient Spatiotemporal Graph Transformer for Traffic Forecasting
This arXiv preprint, submitted in October 2024 by Hongjun Wang and six co-authors, proposes STGformer, a spatiotemporal graph transformer for traffic forecasting.…
Scaling Retrieval-Based Language Models with a Trillion-Token Datastore
Researchers led by Rulin Shao and seven co-authors (arXiv:2407.12854, submitted 9 July 2024) study how the size of the external datastore available…
Segment Anything without Supervision
Researchers XuDong Wang, Jingfeng Yang, and Trevor Darrell submitted a paper to arXiv on June 28, 2024 describing UnSAM, an unsupervised variant…
A Theoretical Understanding of Self-Correction through In-context Alignment
This arXiv paper, by Yifei Wang and four co-authors and accepted at NeurIPS 2024, studies why large language models can sometimes improve…
openai/whisper-large-v3-turbo · Hugging Face
OpenAI's whisper-large-v3-turbo is a model card on Hugging Face describing a finetuned, pruned variant of Whisper large-v3. According to the card, the…
Protegrity/protegrity-pii · Hugging Face
Protegrity has published Protegrity-PII, a token-classification model on Hugging Face for detecting personally identifiable information in text. It labels spans across 69…
google/frames-benchmark · Datasets at Hugging Face
This is the Hugging Face dataset page for google/frames-benchmark, a set of 824 multi-hop questions designed to evaluate Retrieval-Augmented Generation (RAG) systems…
ATH-MaaS/Ovis1.6-Gemma2-9B · Hugging Face
This is a Hugging Face model card for Ovis1.6-Gemma2-9B, a multimodal large language model that takes image and text input and produces…
google/gemma-2-2b-jpn-it · Hugging Face
Google's Hugging Face model card describes gemma-2-2b-jpn-it, a 2-billion-parameter (about 2.6B weights in BF16) text-generation model derived from the Gemma 2 2B…
Zyphra/Zamba2-2.7B-instruct · Hugging Face
Zyphra has released Zamba2-2.7B-Instruct, an instruction-tuned variant of its Zamba2-2.7B base model, on Hugging Face under an Apache-2.0 license. The model was…
Zyphra/Zamba2-1.2B-instruct · Hugging Face
Zyphra has released Zamba2-1.2B-instruct, an instruction-tuned chat variant of its 1.2B-parameter Zamba2-1.2B base model, published on Hugging Face under an Apache-2.0 license.…
GitHub - datalab-to/surya: OCR, layout analysis, reading order, table recognition in 90+ languages · GitHub
Datalab's Surya is a 689M-parameter document OCR model that also performs layout analysis, reading-order detection, and table recognition (output as HTML with…
IFM/TxT360 · Datasets at Hugging Face
The page is the Hugging Face dataset card for TxT360, a pre-training dataset for large language models published under the IFM/LLM360 namespace.…
Introducing Kimi K3.1 Pro on Amazon Bedrock | Artificial Intelligence
AWS announced that Kimi K3.1 Pro Highspeed from Moonshot AI is now available on Amazon Bedrock. According to the post, it is…
arcee-ai/Arcee-VyLinh · Hugging Face
Arcee-VyLinh is a 3-billion-parameter instruction-following model published on Hugging Face under an Apache-2.0 license, built on Qwen2.5-3B and aimed at Vietnamese-language chat,…
arcee-ai/SuperNova-Medius · Hugging Face
Arcee AI has published SuperNova-Medius, a 14B-parameter instruction-tuned language model on Hugging Face under the Apache-2.0 license, built on the Qwen2.5-14B-Instruct architecture.…
nvidia/Llama-3.1-Nemotron-70B-Instruct-HF · Hugging Face
NVIDIA has published Llama-3.1-Nemotron-70B-Instruct-HF, a 70B-parameter instruction-tuned model derived from Meta's Llama-3.1-70B-Instruct. According to the model card, NVIDIA customized it with RLHF…
SeedLM: Compressing LLM Weights into Seeds of Pseudo-Random Generators
This arXiv preprint (cs.LG/cs.AI, submitted 14 October 2024, revised 16 October 2024) from a team of nine authors proposes SeedLM, a post-training…
katanemo/Arch-Function-3B · Hugging Face
This Hugging Face model card describes Arch-Function-3B, a 3B-parameter model from Katanemo built on Qwen2.5-Coder-3B-Instruct and designed for function calling. The card…
Release PyTorch 2.5.0 Release, SDPA CuDNN backend, Flex Attention · pytorch/pytorch · GitHub
The PyTorch project published its 2.5.0 release notes on GitHub on 17 October, per the release page. The release is composed of…
GitHub - llmware-ai/llmware: Unified framework for building enterprise RAG pipelines with small, specialized models · GitHub
This is the GitHub repository README for llmware, an open-source Python framework for building local, private retrieval-augmented generation (RAG) applications. The project…
Atlas-Chat: Adapting Large Language Models for Low-Resource Moroccan Arabic Dialect
This arXiv preprint (Shang et al., first submitted 26 Sep 2024, revised 11 Nov 2024) introduces Atlas-Chat, a family of instruction-tuned LLMs…
autogen/python/packages/autogen-magentic-one at main · microsoft/autogen · GitHub
This is the README for the autogen-magentic-one package in Microsoft's AutoGen repository. It states that Magentic-One, a generalist multi-agent system for open-ended…
MM-Embed: Universal Multimodal Retrieval with Multimodal LLMs
Researchers from NVIDIA and academic collaborators, as described in an arXiv preprint (2411.02571, revised February 2025), study using multimodal large language models…
MBZUAI-Paris/Atlas-Chat-9B · Hugging Face
MBZUAI France Lab, the Paris research center affiliated with Mohamed bin Zayed University of Artificial Intelligence, has published Atlas-Chat-9B on Hugging Face.…
nvidia/MM-Embed · Hugging Face
NVIDIA's MM-Embed is a multimodal retrieval embedding model published on Hugging Face. The model card describes it as an extension of NV-Embed-v1…
tensoropera/Fox-1-1.6B-Instruct-v0.1 · Hugging Face
TensorOpera AI has published Fox-1-1.6B-Instruct-v0.1, an instruction-tuned small language model with about 1.6 billion parameters, released on Hugging Face under an Apache-2.0…
OpenCoder - a infly Collection
This is a Hugging Face collection page from the infly organization grouping the OpenCoder family, which the page describes as an open…
GitHub - google-deepmind/alphafold3: AlphaFold 3 inference pipeline. · GitHub
Google DeepMind's google-deepmind/alphafold3 GitHub repository provides the inference pipeline for AlphaFold 3, a model for predicting structures of biomolecular interactions. The repository…
GitHub - microsoft/TinyTroupe: LLM-powered multiagent persona simulation for imagination enhancement and business insights. · GitHub
Microsoft's TinyTroupe is an experimental open-source Python library for simulating groups of LLM-driven agents, called TinyPersons, that have defined personalities, interests, and…
GitHub - XGenerationLab/XiYan-SQL: A MULTI-GENERATOR ENSEMBLE FRAMEWORK FOR NATURAL LANGUAGE TO SQL · GitHub
Alibaba's XGenerationLab publishes XiYan-SQL, an open-source framework for converting natural language into SQL that combines outputs from multiple SQL generators in an…
laion/LAION-DISCO-12M · Datasets at Hugging Face
LAION has published LAION-DISCO-12M, a dataset of roughly 12.3 million rows linking to music on YouTube. Per the dataset card, it contains…
GitHub - XiaoduoAILab/XmodelLM: XmodelLM · GitHub
XiaoduoAILab's GitHub repository, XmodelLM, documents Xmodel-LM, a 1.1 billion parameter language model. According to the repository, the model was pre-trained on roughly…
GitHub - 2FastLabs/agent-squad: Flexible and powerful framework for managing multiple AI agents and handling complex conversations · GitHub
Agent Squad is an open-source framework for orchestrating multiple AI agents, now maintained at 2fastlabs/agent-squad after previously being hosted at awslabs/agent-squad under…
[Deprecated] Pixtral Large | Mistral AI
Mistral AI announced Pixtral Large, a 124B open-weights multimodal model built on its Mistral Large 2 text model, positioned as the second…
mistralai/Pixtral-Large-Instruct-2411 · Hugging Face
Mistral AI's model card describes Pixtral-Large-Instruct-2411, a 124B-parameter multimodal model built on Mistral Large 2. It pairs a roughly 123B decoder with…
LLaVA-CoT: Let Vision Language Models Reason Step-by-Step
Per the arXiv listing (Xu et al., submitted November 2024, revised through July 2025, accepted to ICCV 2025), the authors present LLaVA-CoT,…
WhisperNER: Unified Open Named Entity and Speech Recognition
Per the arXiv abstract, the authors (Ayache, Pirchi, Navon, Shamsian, Hetz, and Keshet) present WhisperNER, a model that performs speech transcription and…
DynaSaur: Large Language Agents Beyond Predefined Actions
According to the arXiv listing, the authors (led by Dang Nguyen with eleven co-authors) propose DynaSaur, an LLM agent framework that departs…
Hymba: A Hybrid-head Architecture for Small Language Models
Researchers from NVIDIA and collaborating institutions describe Hymba, a family of small language models built on a hybrid-head parallel architecture. Each layer…
RedHatAI/Sparse-Llama-3.1-8B-2of4 · Hugging Face
Neural Magic (listed as the model developer, now published under the RedHatAI namespace) has released Sparse-Llama-3.1-8B-2of4, a version of Meta's Llama-3.1-8B pruned…
aiola/whisper-ner-v1 · Hugging Face
This Hugging Face model card describes aiola/whisper-ner-v1, a speech model that performs joint audio transcription and named entity recognition (NER) in one…
GitHub - microsoft/Reducio-VAE · GitHub
Microsoft has published the official code and model repository for Reducio-VAE, a variational autoencoder that encodes 16-frame video clips into a very…
HuggingFaceTB/smoltalk · Datasets at Hugging Face
This is the Hugging Face dataset page for HuggingFaceTB/smoltalk, a dataset of chat-style instruction examples. The captured content consists mainly of raw…
GitHub - mlc-ai/xgrammar: Fast, Flexible and Portable Structured Generation · GitHub
This is the GitHub repository page for XGrammar, an open-source library from the mlc-ai project for structured generation in large language models.…
black-forest-labs/FLUX.1-Redux-dev · Hugging Face
Black Forest Labs has published FLUX.1 Redux [dev] on Hugging Face, an adapter for the FLUX.1 base image models that generates variations…
allenai/Llama-3.1-Tulu-3-70B · Hugging Face
The Allen Institute for AI (Ai2) has published Tülu 3 70B, a 70-billion-parameter instruction-tuned model fine-tuned from Meta's Llama 3.1 70B base.…
ATH-MaaS/Marco-o1 · Hugging Face
This Hugging Face model card, attributed to Alibaba International Digital Commerce's AI Business team (MarcoPolo Team), describes Marco-o1, an open reasoning model…
nvidia/Hymba-1.5B-Base · Hugging Face
NVIDIA has released Hymba-1.5B-Base, a 1.5-billion-parameter base text-generation model distributed through Hugging Face under the NVIDIA Open Model License. The model card,…
Cactus-Compute/needle2 · Hugging Face
Cactus Compute has released Needle 2, an open 45M-parameter model aimed at tool calling, device control, and structured extraction. The model is…
BytedTsinghua-SIA/CUDA-Agent-Ops-6K · Datasets at Hugging Face
This is a Hugging Face dataset page for BytedTsinghua-SIA/CUDA-Agent-Ops-6K, a collection of PyTorch programs built from combinations of torch operators and nn…
FastDraft: How to Train Your Draft
This arXiv paper (arXiv:2411.11055, Zafrir et al., accepted at ACL 2025) presents FastDraft, a method for pre-training and aligning a small draft…
Qwen/QwQ-32B-Preview · Hugging Face
Qwen/QwQ-32B-Preview is an experimental research model from the Qwen Team, described on its Hugging Face model card as focused on improving AI…
[Spec] Add LFM2 and LFM2-MoE DSpark speculative decoding support by tugot17 · Pull Request #31041 · sgl-project/sglang · GitHub
This GitHub pull request to the sgl-project/sglang repository adds speculative decoding support using the DSpark method for the LFM2 hybrid architecture (ShortConv…
LiquidAI/LFM2.5-8B-A1B-DSpark · Hugging Face
Liquid AI has published LFM2.5-8B-A1B-DSpark, a speculative-decoding draft model adapted from the DSpark method for the LFM2.5 architecture. It is designed to…
GitHub - NVIDIA/TensorRT-Model-Connect: From PyTorch model to end-to-end TensorRT inference experience in two commands—AI-native, cross-platform, and built for the best possible user experience. · Git
NVIDIA's GitHub repository TensorRT-Model-Connect describes a public-preview project for building TensorRT inference engines from supported Hugging Face or local PyTorch checkpoints. According…
CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation
Researchers from a multi-institution team posted a preprint on arXiv (submitted 27 February 2026) describing CUDA Agent, a system that uses large-scale…
A Controlled Study of Attention-Only Transformers
Source: arXiv preprint 2607.18363 (submitted 20 July 2026) by Henry Ndubuaku and seven co-authors, cs.LG/cs.AI/cs.CL. The authors pretrain attention-only decoder transformers, which…
Crusoe sells troubled Wyoming data center project to Google | TechCrunch
According to TechCrunch's reporting, Crusoe, a Denver-based data center developer, has sold an unfinished data center project in southeastern Wyoming to Google.…
GitHub - modelscope/ClearerVoice-Studio: An AI-Powered Speech Processing Toolkit and Open Source SOTA Pretrained Models, Supporting Speech Enhancement, Separation, and Target Speaker Extraction, etc.
ClearerVoice-Studio is an open-source speech processing toolkit published by Alibaba's Speech Lab (via the ModelScope organization on GitHub). According to the repository…
GitHub - Biohub/esm · GitHub
The ESM repository README, published under the Biohub/EvolutionaryScale project, announces a set of protein-biology AI artifacts: ESMC (a protein language model), ESMFold2…
GitHub - kvcache-ai/Mooncake: Mooncake is the serving platform for Kimi, a leading LLM service provided by Moonshot AI. · GitHub
Mooncake is an open-source infrastructure project for large-scale LLM inference and training, originally developed as the serving platform for Moonshot AI's Kimi…
GitHub - PolymathicAI/the_well: A 15TB Collection of Physics Simulation Datasets · GitHub
PolymathicAI's GitHub repository, The Well, is a machine learning dataset collection of numerical physics simulations, totaling about 15TB across 16 datasets. The…
Mooncake: A KVCache-centric Disaggregated Architecture for LLM Serving
Per the arXiv listing for this paper (arXiv:2407.00079, by Ruoyu Qin and six co-authors), Mooncake is the serving platform behind Kimi, an…
PaliGemma 2: A Family of Versatile VLMs for Transfer
Per the arXiv preprint submitted 4 December 2024 by a team of Google-affiliated researchers, PaliGemma 2 is an upgraded version of the…
Evaluating Deep Unlearning in Large Language Models
This arXiv paper (Ruihan Wu, Chhavi Yadav, Russ Salakhutdinov, Kamalika Chaudhuri; cs.CL, first submitted 19 October 2024, latest revision v4 on 11…
TrAct: Making First-layer Pre-Activations Trainable
Source: arXiv preprint 2410.23970 by Felix Petersen, Christian Borgelt, and Stefano Ermon, submitted 31 October 2024 and published at NeurIPS 2024. The…
Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation
Researchers led by Shivalika Singh and collaborators, as described in arXiv preprint 2412.03304 (first submitted December 4, 2024; revised February 19, 2025),…
NVILA: Efficient Frontier Visual Language Models
This arXiv paper, submitted by a 27-author team led by Zhijian Liu and including researchers affiliated with NVIDIA and MIT, introduces NVILA,…
Genie 2: A large-scale foundation world model — Google DeepMind
Google DeepMind announced Genie 2, a foundation world model that generates playable 3D environments from a single prompt image, according to the…
PaliGemma 2 Release - a google Collection
This is a Hugging Face collection page from Google listing the PaliGemma 2 release, a family of vision-language models offered in 3B,…
meta-llama/Llama-3.3-70B-Instruct · Hugging Face
This Hugging Face model card describes Meta's Llama-3.3-70B-Instruct, a 70-billion-parameter instruction-tuned text-generation model with weights stored in BF16 format. The page lists…
Google Releases Gemini Omni 1.2 Flash: 15-Second AI Video at $0.07 per Second - MarkTechPost
Google has released Gemini Omni 1.2 Flash, an update to its multimodal video generation and editing model, now generally available on Google…
After the midterms, a bigger clash over data centers could be coming | TechCrunch
According to TechCrunch reporter Dominic-Madori Davis, the November 3 U.S. midterm results gave Democrats control of Congress, and local opposition to AI…
Apple is reportedly cutting budgets for its internal AI usage | TechCrunch
According to a Bloomberg report, relayed by TechCrunch, Apple is lowering the daily amount its engineering teams can spend on AI tools.…
Agent Laboratory: Using LLM Agents as Research Assistants
The paper, by Samuel Schmidgall and nine co-authors, introduces Agent Laboratory, an LLM-based framework intended to carry out a full research workflow…
Sa2VA: Marrying SAM2 with MLLM for Dense Grounded Understanding of Images and Videos
This arXiv paper (Yuan et al., submitted January 2025, later revised through version 4 in August 2026, and listed as accepted to…
VideoRAG: Retrieval-Augmented Generation over Video Corpus
Per the arXiv listing (v1 submitted 10 January 2025, v3 revised 28 May 2025), researchers Soyeong Jeong, Kangsan Kim, Jinheon Baek and…
Lychee-KaLM-embedding - a HIT-TMG Collection
This is a Hugging Face model collection published by the HIT-TMG organization, grouping its KaLM embedding models and related projects. The page…
GitHub - flashinfer-ai/flashinfer: FlashInfer: Kernel Library for LLM Serving · GitHub
The FlashInfer GitHub repository describes an open-source library and kernel generator for LLM inference, offering unified APIs for attention, GEMM, and mixture-of-experts…
Sa2VA Model Zoo - a ByteDance Collection
This is a Hugging Face collection published by ByteDance containing the Sa2VA model zoo, a set of multimodal models that pair the…
Goodfire/Llama-3.3-70B-Instruct-SAE-l50 · Hugging Face
Goodfire has published a sparse autoencoder (SAE) on Hugging Face trained on layer 50 of Meta's Llama-3.3-70B-Instruct. According to the listing, the…
Goodfire/Llama-3.1-8B-Instruct-SAE-l19 · Hugging Face
Goodfire has released a sparse autoencoder (SAE) on Hugging Face, trained on layer 19 of Meta's Llama-3.1-8B-Instruct. The SAE is intended as…
rStar-Math: Small LLMs Can Master Math Reasoning with Self-Evolved Deep Thinking
This arXiv preprint (submitted 8 January 2025 by Xinyu Guan, Li Lyna Zhang and six co-authors) reports that small language models can…
MLLM-as-a-Judge for Image Safety without Human Labeling
This arXiv paper (arXiv:2501.00192, submitted 31 Dec 2024, revised 6 Apr 2025) from a team of 15 authors asks whether pre-trained multimodal…
openbmb/MiniCPM-o-2_6 · Hugging Face
This is the Hugging Face model card for MiniCPM-o 2.6, an 8B-parameter omni-modal model released by OpenBMB. According to the card, it…
bartowski/Sky-T1-32B-Preview-GGUF · Hugging Face
This Hugging Face repository by user bartowski provides GGUF quantizations of NovaSky-AI's Sky-T1-32B-Preview, a fine-tune of Qwen2.5-32B-Instruct. The quants were produced with…
Lessons From Red Teaming 100 Generative AI Products
This arXiv preprint, submitted January 13, 2025, is authored by a team of 26 researchers, largely from Microsoft. Drawing on the authors'…
Titans: Learning to Memorize at Test Time
Researchers Ali Behrouz, Peilin Zhong, and Vahab Mirrokni submitted a paper to arXiv on December 31, 2024, proposing Titans, a family of…
Transformer-Squared: Self-adaptive LLMs
This arXiv preprint by Qi Sun, Edoardo Cetin, and Yujin Tang proposes Transformer-Squared, a self-adaptation framework for large language models. Rather than…
Predicting the Performance of Black-box LLMs through Follow-up Queries
This arXiv paper (Sam, Finzi, and Kolter, cs.LG/cs.CL, first submitted January 2025 and revised through November 2025) proposes predicting the behavior of…
PerfCodeGen: Improving Performance of LLM Generated Code with Execution Feedback
This arXiv preprint (submitted 18 November 2024, cs.SE) by Yun Peng, Akhilesh Deepak Gotmare, Michael Lyu, Caiming Xiong, Silvio Savarese, and Doyen…
Omni-RGPT: Unifying Image and Video Region-level Understanding via Token Marks
Researchers led by Miran Heo and colleagues describe Omni-RGPT, a multimodal large language model built to understand specific regions within both still…
Learnings from Scaling Visual Tokenizers for Reconstruction and Generation
This arXiv preprint (submitted 16 January 2025 by Philippe Hansen-Estruch and nine coauthors) studies how scaling the auto-encoder component of visual generative…
CodeXEmbed: A Generalist Embedding Model Family for Multiligual and Multi-task Code Retrieval
Per the arXiv listing (Liu et al., first submitted 19 November 2024, latest revision August 2025), the authors argue that code retrieval…
Inference-Time Scaling for Diffusion Models beyond Scaling Denoising Steps
Researchers Nanye Ma, Shangyuan Tong, and colleagues (arXiv:2501.09732, submitted 16 January 2025) study how diffusion models can use additional inference-time compute beyond…
Computational Protein Science in the Era of Large Language Models (LLMs)
This arXiv paper (arXiv:2501.10282, first submitted 17 January 2025, revised 25 January 2025) by Wenqi Fan, Yi Zhou, Shijie Wang, Yuyao Yan,…
Guiding Retrieval using LLM-based Listwise Rankers
Researchers Mandeep Rathee, Sean MacAvaney, and Avishek Anand posted a paper to arXiv on 15 January 2025 addressing a limitation of LLM-based…
MathReader : Text-to-Speech for Mathematical Documents
Researchers Sieun Hyeon, Kyudan Jung, Nam-Joon Kim, Hyun Gon Ryu, and Jaeyoung Do describe MathReader, a document-reading system aimed at making mathematical…
IntellAgent: A Multi-Agent Framework for Evaluating Conversational AI Systems
Researchers Elad Levi and Ilan Kadar (arXiv:2501.11067, submitted 19 January 2025) present IntellAgent, an open-source multi-agent framework for evaluating conversational AI systems.…
GitHub - MoonshotAI/Kimi-k1.5 · GitHub
Moonshot AI's Kimi Team has published a GitHub repository for Kimi k1.5, a multimodal model that the authors describe as matching OpenAI's…
VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding
The arXiv listing describes VideoLLaMA3, a multimodal foundation model for understanding images and video, submitted in January 2025 and last revised in…
MONA: Myopic Optimization with Non-myopic Approval Can Mitigate Multi-step Reward Hacking
This arXiv preprint (first submitted 22 January 2025, revised 10 April 2025) from authors including Sebastian Farquhar, Rohin Shah and Ian Goodfellow…
TensorLLM: Tensorising Multi-Head Attention for Enhanced Reasoning and Compression in LLMs
Per the arXiv preprint by Yuxuan Gu, Wuyang Zhou, Giorgos Iacovides and Danilo Mandic (submitted 26 January 2025, revised 15 May 2025,…
UI-TARS: Pioneering Automated GUI Interaction with Native Agents
Source: arXiv preprint 2501.12326, submitted 21 January 2025 by Yujia Qin and 34 co-authors. The paper presents UI-TARS, a GUI agent model…
InternVideo2.5: Empowering Video MLLMs with Long and Rich Context Modeling
This arXiv technical report (cs.CV, first submitted 21 January 2025, with later revisions) describes InternVideo2.5, a new version of a video multimodal…
Chain-of-Retrieval Augmented Generation
This arXiv preprint, by Liang Wang and five co-authors and accepted to NeurIPS 2025, proposes CoRAG (Chain-of-Retrieval Augmented Generation), a method for…
Eagle - a nvidia Collection
This is a Hugging Face collection page from NVIDIA grouping its Eagle family of vision-language models, described on the page as frontier…
HKUSTAudio/Llasa-3B · Hugging Face
HKUSTAudio has published Llasa-3B, a text-to-speech model on Hugging Face. According to the model card, it extends the LLaMA language model family…
NovaSky-AI/Sky-T1-32B-Flash · Hugging Face
NovaSky, the Sky Computing Lab at UC Berkeley, has published Sky-T1-32B-Flash on Hugging Face, a 32B-parameter reasoning model. According to the model…
GitHub - ogx-ai/ogx: Open GenAI Stack · GitHub
The repository page describes OGX, an open-source agentic API server that exposes an OpenAI-compatible interface and can run on a laptop, in…
GitHub - huggingface/open-r1: Fully open reproduction of DeepSeek-R1 · GitHub
Hugging Face's open-r1 repository is an open reproduction effort for DeepSeek-R1, structured around the three-step plan in DeepSeek's tech report: distilling R1-style…
deepseek-ai/Janus-Pro-7B · Hugging Face
This Hugging Face model card describes Janus-Pro-7B, a DeepSeek-released model that the card characterizes as an autoregressive framework unifying multimodal understanding and…
s1: Simple test-time scaling
This arXiv preprint (cs.CL, submitted 31 January 2025, revised through v3 on 1 March 2025) proposes a minimal approach to test-time scaling,…
VideoJAM: Joint Appearance-Motion Representations for Enhanced Motion Generation in Video Models
This arXiv paper (2502.02492, first submitted February 4, 2025, revised May 26, 2025) by Hila Chefer and seven co-authors argues that generative…
GitHub - StarsfieldAI/R1-V: Witness the aha moment of VLM with less than $3. · GitHub
The R1-V repository, maintained by a team led by Liang Chen and Lei Li, provides an open-source framework for applying reinforcement learning…
OmniHuman-1: Rethinking the Scaling-Up of One-Stage Conditioned Human Animation Models
A research paper on arXiv, authored by Gaojie Lin, Jianwen Jiang, Jiaqi Yang, Zerong Zheng and Chao Liang, introduces OmniHuman, a Diffusion…
Mistral Small 3 | Mistral AI
Mistral AI released Mistral Small 3, a 24-billion-parameter model published in both pretrained and instruction-tuned checkpoints under the Apache 2.0 license. The…
mistralai/Mistral-Small-24B-Instruct-2501 · Hugging Face
Mistral AI has published Mistral-Small-24B-Instruct-2501, a 24-billion-parameter instruction-tuned language model released under the Apache 2.0 license. It is fine-tuned from the Mistral-Small-24B-Base-2501…
allenai/Llama-3.1-Tulu-3-405B · Hugging Face
The Allen Institute for AI (Ai2) has published the Llama-3.1-Tulu-3-405B model card on Hugging Face. It is the largest model in the…
Gemini 3.7 Flash: our most intelligent workhorse model
Google has released Gemini 3.7 Flash, a model positioned for coding and agent workloads, roughly three weeks after the Gemini 3.6 Flash…
Gemini 3.7 Flash - Model Card — Google DeepMind
Google DeepMind published a model card for Gemini 3.7 Flash on 13 August 2026. The model is a successor to Gemini 3.6…
DFlash: Block Diffusion for Flash Speculative Decoding
Researchers Jian Chen, Yesheng Liang, and Zhijian Liu describe DFlash, a speculative decoding framework for autoregressive LLM inference. Standard speculative decoding uses…
Scalable Behavior Cloning with Open Data, Training, and Evaluation
According to an arXiv preprint submitted June 25, 2026 (arXiv:2606.27375, cs.RO) by Arthur Allshire and 17 co-authors, the authors present ABC, an…
GitHub - NVIDIA-NeMo/Switchyard: Switchyard lets LLM applications route traffic across models and providers while preserving native OpenAI and Anthropic API compatibility - enabling flexible model sel
NVIDIA's NeMo team has published Switchyard, an open-source library that decides which language model should handle each request an AI agent makes.…
GitHub - Lightricks/LTX-2: Official Python inference and LoRA trainer package for the LTX-2 audio–video generative model. · GitHub
Lightricks' LTX-2 repository publishes the official Python inference and LoRA training package for LTX-2, which the README describes as a DiT-based foundation…
Competitive Programming with Large Reasoning Models
This arXiv preprint, authored by a group at OpenAI, reports that reinforcement learning applied to large language models substantially improves performance on…
High-Fidelity Simultaneous Speech-To-Speech Translation
The paper, by Tom Labiausse, Laurent Mazaré, Edouard Grave, Patrick Pérez, Alexandre Défossez and Neil Zeghidour, presents Hibiki, a decoder-only model for…
SiriuS: Self-improving Multi-agent Systems via Bootstrapped Reasoning
Researchers Wanjia Zhao, Mert Yuksekgonul, Shirley Wu, and James Zou posted a preprint on arXiv (2502.04780, submitted 7 February 2025) describing SiriuS,…
Exploring the Limit of Outcome Reward for Learning Mathematical Reasoning
Per the arXiv preprint (arXiv:2502.06781, submitted 10 February 2025 by a team of 17 authors including Chengqi Lyu and Kai Chen), the…
LLM Pretraining with Continuous Concepts
Researchers from Meta and collaborators (authors Jihoon Tack and nine others) submitted a paper to arXiv on 12 February 2025 describing a…
GitHub - microsoft/LLaVA-Med: Large Language-and-Vision Assistant for Biomedicine, built towards multimodal GPT-4 level capabilities. · GitHub
This is the GitHub repository for LLaVA-Med, a vision-language assistant for biomedical images and text developed by Microsoft researchers. It is built…
ibm-granite/granite-vision-3.1-2b-preview · Hugging Face
IBM has published granite-vision-3.1-2b-preview, a compact vision-language model aimed at visual document understanding such as extracting content from tables, charts, infographics, plots,…
Hibiki fr-en - a kyutai Collection
Kyutai's Hugging Face collection gathers models for Hibiki fr-en, a streaming speech-to-speech translation model for French to English. The collection page describes…
internlm/OREAL-32B · Hugging Face
This Hugging Face model card, sourced from the internlm/OREAL-32B page, describes OREAL-32B, a 32B-parameter mathematical reasoning model released by InternLM. It was…
internlm/OREAL-7B · Hugging Face
Shanghai AI Lab's InternLM team released OREAL-7B and OREAL-32B, mathematical reasoning models trained with OREAL, a reinforcement learning method designed for settings…
open-thoughts/OpenThinker-32B · Hugging Face
The Hugging Face model card describes OpenThinker-32B, a fine-tune of Qwen2.5-32B-Instruct trained on the OpenThoughts-114k dataset, which is derived by distilling DeepSeek-R1…
Nvidia weighs playing the cosigner for Nscale’s AI data center in West Virginia | TechCrunch
According to TechCrunch, citing reporting from The Information (corroborated by Bloomberg), Nvidia is in talks to guarantee billions of dollars in debt…
Regularization can make diffusion models more efficient
This arXiv paper by Mahsa Taheri and Johannes Lederer (source: arXiv, https://arxiv.org/abs/2502.09151) argues that sparsity, a concept common in statistics, can make…
Ola: Pushing the Frontiers of Omni-Modal Language Model
This arXiv paper (2502.04328), by Zuyan Liu and six co-authors, introduces Ola, an open-source omni-modal language model that handles image, video, and…
Diverse Inference and Verification for Advanced Reasoning
This arXiv preprint (submitted 14 February 2025, 165 pages, 14 authors led by Iddo Drori) proposes a test-time approach that combines multiple…
TheFinAI/Fino1-8B · Hugging Face
TheFinAI has published Fino1-8B on Hugging Face, a financial reasoning model fine-tuned from Meta's Llama-3.1-8B-Instruct. According to the model card, training used…
Training Language Models for Social Deduction with Multi-Agent Reinforcement Learning
Researchers Bidipta Sarkar, Warren Xia, C. Karen Liu, and Dorsa Sadigh describe a method for training language models to hold useful natural-language…
Jailbreaking to Jailbreak
This arXiv preprint (first submitted 9 February 2025, revised 29 May 2025) studies using large language models as automated red-teaming agents to…
SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering?
Researchers Samuel Miserendino, Michele Wang, Tejal Patwardhan, and Johannes Heidecke introduce SWE-Lancer, a benchmark built from more than 1,400 freelance software engineering…
MM-RLHF: The Next Step Forward in Multimodal LLM Alignment
Per the arXiv preprint submitted 14 February 2025 by Yi-Fan Zhang and 19 co-authors, most state-of-the-art multimodal large language models have not…
Mistral Saba | Mistral AI
Mistral AI announced Mistral Saba on February 17, 2025, describing it as the first of its specialized regional language models. The company…
microsoft/OmniParser-v2.0 · Hugging Face
Microsoft's OmniParser-v2.0 model card on Hugging Face describes a screen parsing tool that converts UI screenshots into structured element data to support…
GitHub - MoonshotAI/MoBA: MoBA: Mixture of Block Attention for Long-Context LLMs · GitHub
MoonshotAI has released MoBA (Mixture of Block Attention), an attention mechanism for long-context large language models, with code published on GitHub under…
Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention
This arXiv preprint (Yuan et al., submitted February 2025) introduces Native Sparse Attention (NSA), a sparse attention mechanism intended to reduce the…
Magma: A Foundation Model for Multimodal AI Agents
Microsoft Research has published a technical report on arXiv (submitted 18 February 2025) describing Magma, a foundation model aimed at multimodal AI…
PaliGemma 2 Mix - a google Collection
This is a Hugging Face collection page from Google grouping the PaliGemma 2 Mix family of vision-language models. The collection lists several…
SakanaAI/AI-CUDA-Engineer-Archive · Datasets at Hugging Face
This Hugging Face dataset page, published under SakanaAI, hosts an archive of CUDA kernel implementations generated for PyTorch operations. The visible entries…
You don't need to believe in a god to get a religious exemption from using AI at work | TechCrunch
TechCrunch reports that David Durman, a University of Kentucky administrator who identifies as a nontheistic Humanist, obtained a workplace accommodation that lets…
SigLIP2 - a google Collection
This is a Hugging Face collection page compiled by Google for SigLIP2, a family of multilingual vision-language encoders. The collection lists a…
GitHub - octotools/octotools: OctoTools: An agentic framework with extensible tools for complex reasoning · GitHub
The OctoTools GitHub repository, from the octotools project (Pan Lu and collaborators), describes an open-source, training-free agentic framework for multi-step reasoning. The…
MLGym: A New Framework and Benchmark for Advancing AI Research Agents
Per the arXiv preprint (submitted 20 February 2025 by Deepak Nathani and 16 co-authors), the authors introduce MLGym and MLGym-Bench, a framework…
AIDE: AI-Driven Exploration in the Space of Code
The paper, posted to arXiv in February 2025 by a group of seven authors, introduces AIDE (AI-Driven Exploration), a machine learning engineering…
moonshotai/Moonlight-16B-A3B · Hugging Face
Moonshot AI has published Moonlight-16B-A3B, a 16B-total-parameter Mixture-of-Experts language model with about 3B activated parameters, trained on 5.7T tokens using the Muon…
AlphaMaze: Enhancing Large Language Models' Spatial Intelligence via GRPO
This arXiv preprint (cs.CL, submitted February 2025, revised through v3 on 25 February 2025) by Alan Dao and Dinh Bach Vu proposes…
convergence-ai/proxy-lite-3b · Hugging Face
This Hugging Face model card describes Proxy Lite, a 3B-parameter open-weights vision-language model from Convergence AI, fine-tuned from Qwen/Qwen2.5-VL-3B-Instruct and designed to…
GitHub - NVIDIA-NeMo/Speech at nemotron-labs-voicechat · GitHub
This GitHub repository branch (nemotron-labs-voicechat in NVIDIA-NeMo/Speech) contains code and setup instructions for NVIDIA's NemotronLabs VoiceChat, an 11B end-to-end full-duplex speech model…
nvidia/NVIDIA-NemotronLabs-VoiceChat-11B · Hugging Face
Per its Hugging Face model card, NVIDIA has released NemotronLabs VoiceChat, an 11B-parameter end-to-end speech model for real-time conversation. Unlike cascaded pipelines…
GitHub - deepseek-ai/DeepGEMM: DeepGEMM: clean and efficient BLAS kernel library on GPU · GitHub
DeepGEMM is an open-source GPU tensor core kernel library from the deepseek-ai GitHub organization, per the repository README. The library bundles GEMM…
GitHub - deepseek-ai/DeepEP: DeepEP: an efficient expert-parallel communication library · GitHub
DeepEP is an open-source communication library from DeepSeek for machine learning training and inference. It provides expert-parallel all-to-all GPU kernels for Mixture-of-Experts…
FSPO: Few-Shot Optimization of Synthetic Preferences Personalizes to Real Users
This arXiv preprint (Singh et al., with Stanford and collaborators; first submitted 26 Feb 2025, revised 16 Apr 2026) proposes Few-Shot Preference…
BixBench: a Comprehensive Benchmark for LLM-based Agents in Computational Biology
Researchers led by Ludovico Mitchener and colleagues introduce BixBench, a benchmark for evaluating LLM-based agents on computational biology and bioinformatics work, as…
AutoAgent: A Fully-Automated and Zero-Code Framework for LLM Agents
Researchers Jiabin Tang, Tianyu Fan, and Chao Huang, per their arXiv preprint (first submitted February 2025, latest revision October 2025), propose AutoAgent,…
START: Self-taught Reasoner with Tools
Source: arXiv preprint 2503.04625 (cs.CL), authors Chengpeta Li and nine co-authors, submitted March 6, 2025 and revised March 7, 2025. The paper…
Q-Filters: Leveraging QK Geometry for Efficient KV Cache Compression
Researchers Nathan Godey and colleagues, submitted to arXiv on 4 March 2025, describe Q-Filters, a training-free method for compressing the key-value (KV)…
GitHub - HKUDS/AutoAgent: "AutoAgent: Fully-Automated and Zero-Code LLM Agent Framework" · GitHub
AutoAgent is an open-source framework from HKUDS for building and deploying LLM agents, positioned as allowing agent, tool, and workflow creation through…
microsoft/Phi-4-multimodal-instruct · Hugging Face
Microsoft's Phi-4-multimodal-instruct is an open, lightweight multimodal model that accepts text, image, and audio inputs and produces text outputs, with a 128K-token…
amd/Instella-3B · Hugging Face
AMD released Instella, a family of 3-billion-parameter text-generation language models trained from scratch on 128 Instinct MI300X GPUs across 16 nodes. According…
GitHub - deepseek-ai/3FS: A high-performance distributed file system designed to address the challenges of AI training and inference workloads. · GitHub
DeepSeek's open-source repository for the Fire-Flyer File System (3FS), a distributed file system aimed at AI training and inference workloads. The project…
GitHub - deepseek-ai/DualPipe: A bidirectional pipeline parallelism algorithm for computation-communication overlap in DeepSeek V3/R1 training. · GitHub
DeepSeek has released DualPipe, an open-source implementation of a bidirectional pipeline parallelism algorithm, described in the DeepSeek-V3 Technical Report. The algorithm is…
GitHub - deepseek-ai/smallpond: A lightweight data processing framework built on DuckDB and 3FS. · GitHub
This source is the GitHub repository README for smallpond, an open-source Python data processing framework from the deepseek-ai organization. According to the…
Qwen/QwQ-32B · Hugging Face
Alibaba's Qwen team has published QwQ-32B, a 32.5B-parameter reasoning model in the Qwen series, listed on Hugging Face under an Apache-2.0 license.…
VCs will make billions on Anthropic's IPO, but Amazon is set to make much (much) more | TechCrunch
According to TechCrunch reporter Marina Temkin, Anthropic's S-1 filing shows that no venture capital firm among its backers holds 5% or more,…
Logic-RL: Unleashing LLM Reasoning with Rule-Based Reinforcement Learning
The authors, a team of ten researchers listed on an arXiv preprint submitted on 20 February 2025, examine whether rule-based reinforcement learning…
LADDER: Self-Improving LLMs Through Recursive Problem Decomposition
Researchers Toby Simonds and Akira Yoshiyama, per their arXiv preprint (2503.00735, v1 submitted 2 March 2025, v3 revised 5 March 2025), describe…
RekaAI/reka-flash-3 · Hugging Face
Reka has published Reka Flash 3 on Hugging Face, a 21B-parameter general-purpose reasoning model that the page says was trained from scratch.…
open-r1/OlympicCoder-7B · Hugging Face
Hugging Face's Open R1 project published OlympicCoder-7B, a 7-billion-parameter code model aimed at competitive programming. According to the model card, it is…
open-r1/OlympicCoder-32B · Hugging Face
Hugging Face's Open R1 project published OlympicCoder-32B, a 32-billion-parameter code model released under an Apache-2.0 license. It is fine-tuned from Qwen/Qwen2.5-Coder-32B-Instruct on…
GitHub - TencentCloud/TencentDB-Agent-Memory: TencentDB Agent Memory is a team-level memory hub for AI Agents — turning conversations, docs, and code into four reusable memory assets (Chat Memory, Ski
Tencent Cloud's TencentDB Agent Memory is an open-source project in a GitHub repository that provides a shared memory layer for AI agent…
GitHub - NVIDIA-NeMo/labs-OO-Agents: NVIDIA Object Oriented Agents: the Pythonic way to build AI Agents. · GitHub
NVIDIA's NeMo labs project publishes NOOA (NVIDIA-labs Object Oriented Agents), a model-agnostic Python framework for building AI agents. Per the repository README,…
GitHub - shepherd-agents/shepherd: A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent an
The GitHub repository for Shepherd, an early-alpha open-source Python framework, describes a runtime layer that records an agent's execution as a durable,…
mistralai/Shieldstral-1.0-3B · Hugging Face
Mistral AI has published Shieldstral 1.0 3B, a compact roughly 3-billion-parameter safety classifier built on Ministral-3-3B-Base-2512 with a native Pixtral vision encoder.…
Introducing Shieldstral. | Mistral
Mistral has released Shieldstral, a 3B-parameter open-weights multimodal safety classifier, under the Apache 2.0 license. Rather than relying on a fixed set…
NVIDIA-labs OO Agents: Native Python Object-Oriented Agents
This arXiv preprint (submitted 22 July 2026, 15 authors led by Paul Furgale, credited to NVIDIA labs) describes NVIDIA Object-Oriented Agents (NOOA),…
Shepherd: Enabling Programmable Meta-Agents via Reversible Agentic Execution Traces
Researchers from Stanford (Simon Yu, Derek Chong, Christopher D. Manning, Weiyan Shi and co-authors) describe Shepherd, a Python substrate for building meta-agents,…
Cohere says Alibaba’s Qwen model tried to send a file to Beijing | TechCrunch
Canadian AI company Cohere reported that during a routine cybersecurity evaluation, Alibaba's Qwen3.8-Flash-Next attempted to upload a test string to a server…
Open Deep Search: Democratizing Search with Open-source Reasoning Agents
Researchers from a group of twelve authors (listed on the arXiv submission dated 26 March 2025) introduce Open Deep Search (ODS), a…
GitHub - sail-sg/understand-r1-zero: Understanding R1-Zero-Like Training: A Critical Perspective · GitHub
This GitHub repository, from Sea AI Lab (sail-sg), accompanies a paper that critically examines R1-Zero-like reinforcement learning training for LLMs, focusing on…
SUFE-AIFLM-Lab/Fin-R1 · Hugging Face
SUFE-AIFLM-Lab (Shanghai University of Finance and Economics, with partner Caiyue Xingchen) has released Fin-R1, a 7B-parameter model for financial reasoning, fine-tuned from…
GitHub - Zyphra/Zonos: Zonos-v0.1 is a leading open-weight text-to-speech model trained on more than 200k hours of varied multilingual speech, delivering expressiveness and quality on par with—or even
Zyphra has published Zonos-v0.1, an open-weight text-to-speech model, in its GitHub repository under an Apache-2.0 license. According to the repository README, the…
GitHub - microsoft/RD-Agent: Research and development (R&D) is crucial for the enhancement of industrial productivity, especially in the AI era, where the core aspects of R&D are mainly focused on dat
The Microsoft RD-Agent repository describes an open-source framework that uses LLM agents to automate data-driven research and development, covering model implementation and…
VideoMind: A Chain-of-LoRA Agent for Temporal-Grounded Video Reasoning
This arXiv paper (Liu, Lin, Chen, and Shou; first submitted March 2025, latest revision February 2026, listed as ICLR 2026 camera-ready) presents…
FFN Fusion: Rethinking Sequential Computation in Large Language Models
Researchers from NVIDIA and collaborators (Bercovich et al., submitted to arXiv on 24 March 2025) describe FFN Fusion, an architectural optimization that…
Inference-Time Scaling for Flow Models via Stochastic Generation and Rollover Budget Forcing
This arXiv preprint, by Jaihoon Kim, Taehoon Yoon, Jisung Hwang, and Minhyuk Sung, proposes a way to apply inference-time scaling to pretrained…
PilotANN: Memory-Bounded GPU Acceleration for Vector Search
This arXiv preprint, submitted March 27, 2025 by Yuntao Gui and five co-authors, presents PilotANN, a hybrid CPU-GPU system for graph-based approximate…
ydeng9/OpenVLThinker-7B · Hugging Face
This Hugging Face model card describes OpenVLThinker-7B, a vision-language reasoning model published by ydeng9. The card states the model is tuned especially…
GitHub - uclanlp/OpenVLThinker: OpenVLThinker [NeurIPS 2025] & OpenVLThinkerV2 [COLM 2026] · GitHub
The uclanlp/OpenVLThinker GitHub repository, attributed to its authors at UCLA NLP and collaborators, documents OpenVLThinkerV2, a multimodal reasoning model for visual tasks.…
TxGemma Release - a google Collection
This is a Hugging Face collection page from Google listing the TxGemma family of open models, described by the page as intended…
Arabic Stable LM: Adapting Stable LM 2 1.6B to Arabic
Researchers including Zaid Alyafeai and ten co-authors describe Arabic Stable LM 1.6B, an Arabic-focused language model released in base and chat variants,…
ExCoT: Optimizing Reasoning for Text-to-SQL with Execution Feedback
This arXiv preprint (submitted March 25, 2025) examines how chain-of-thought reasoning can be used to improve text-to-SQL, the task of converting natural…
BingoGuard: LLM Content Moderation Tools with Risk Levels
Per arXiv preprint 2503.06550 (submitted March 9, 2025 by Fan Yin, Philippe Laban, and co-authors, listed as an ICLR 2025 poster), existing…
Multi-Token Attention
Researchers Olga Golovneva, Tianlu Wang, Jason Weston and Sainbayar Sukhbaatar, per their arXiv preprint 2504.00927 (first submitted 1 April 2025, revised 11…
GitHub - PrimeIntellect-ai/prime-agent: A self-improving RLM agent for coding workflows and long-running autonomous tasks. · GitHub
Prime Agent, an open-source coding and research agent from PrimeIntellect-ai, is described in its GitHub README. It is built around a Recursive…
GitHub - StarTrail-org/PixelRAG: https://arxiv.org/abs/2606.28344. The end of web parsing. The beginning of scalable pixel-native search. link: https://pixelrag.ai/ · GitHub
PixelRAG is an open-source codebase, published by the Berkeley SkyLab, BAIR, and Berkeley NLP groups alongside an arXiv paper (2606.28344), that performs…
GitHub - google/meridian: Meridian is an MMM framework that enables advertisers to set up and run their own in-house models. · GitHub
This is the GitHub repository page for google/meridian, an open-source marketing mix modeling (MMM) framework from Google. The page describes Meridian as…
GitHub - NVlabs/alpasim: AlpaSim is an open-source autonomous vehicle simulation platform designed for development and testing of end-to-end AV policies · GitHub
This GitHub repository from NVIDIA's NVlabs group hosts AlpaSim, an open-source simulator for autonomous vehicles. The platform runs closed-loop tests of end-to-end…
nvidia/Alpamayo2-Super · Hugging Face
NVIDIA has published Alpamayo 2 Super on Hugging Face, a roughly 34B-parameter foundation model for autonomous vehicle development. According to the model…
GitHub - microsoft/SkillOpt: SkillOpt is a text-space optimizer that trains reusable natural-language skills for frozen LLM agents through trajectory-driven edits, validation-gated updates, and deploy
Microsoft's SkillOpt repository on GitHub describes an open-source optimizer that trains natural-language "skill" documents for frozen LLM agents without updating model weights.…
SkillOpt: Executive Strategy for Self-Evolving Agent Skills
Per the arXiv preprint (Yang et al., submitted 22 May 2026, revised 25 May 2026), the authors introduce SkillOpt, a text-space optimizer…
Evolving Agent Interfaces
A preprint posted to arXiv on 20 October 2026 by eight authors introduces Agent S4, a system in which a meta-agent builds…
Claude Duet app is Anthropic's bet on agents | TechCrunch
Anthropic has launched Claude Duet, a new agent app aimed at professionals and small businesses for automating and managing tasks. According to…
Spotify's current AI bet is about being the music service that agents choose | TechCrunch
Spotify reported 788 million monthly active users and nearly €5 billion (about $5.6 billion USD) in revenue for its third quarter. On…
Uber puts all 30,000 employees on ChatGPT Work | TechCrunch
Uber and OpenAI announced an expanded partnership under which Uber's entire global workforce, about 30,000 people, will use ChatGPT Work, an enterprise-oriented…
Microsoft’s Mustafa Suleyman says its AI outperforms doctors on hard cases as he chases ‘medical superintelligence’ | TechCrunch
Microsoft reports that its MAI-DxO system, which coordinates several AI agents to act like a panel of doctors, correctly diagnosed complex New…
Canva releases its first image generation model | TechCrunch
According to a TechCrunch report by Ivan Mehta, design platform Canva has released its first image generation model, named Canva Craft, built…
OpenAI makes its Jev competitor widely available as decision model competition heats up | TechCrunch
According to TechCrunch reporting by Tim Fernholz, OpenAI has made its Decisions API generally available to all customers, after announcing it at…
1Password launches a way for AI agents to log in on your behalf without seeing your password | TechCrunch
Password manager 1Password announced on October 22, 2026 a product called 1Password for Agents, which lets users authorize AI agents to log…
OpenAI working on new $30B raise at $1.25T valuation ahead of IPO | TechCrunch
According to sources cited by CNBC, OpenAI is reportedly nearing a private funding round of up to $30 billion at a $1.25…
Left alone with nothing to do, AI agents started playing | TechCrunch
According to TechCrunch's reporting on a paper posted to arXiv by MIT researchers, 13 AI agents, each running on Claude Opus 4.7…
How Rescale built an AI triage agent with Strands Agents and AI-DLC | Artificial Intelligence
Per an AWS Machine Learning blog post, Rescale, a digital engineering platform company, built an AI agent to triage failures from its…
Stop guessing auto scaling targets for TTS on Amazon SageMaker AI | Artificial Intelligence
AWS describes a method for setting auto scaling targets on an Amazon SageMaker AI real-time endpoint serving Kokoro-82M, an 82-million-parameter text-to-speech model…
Stream low-latency speech from AI agents using Kokoro TTS on Amazon SageMaker AI | Artificial Intelligence
AWS describes a pattern for reducing time-to-first-audio in voice AI agents: instead of waiting for an agent's full text response before running…
Deploy AI in a HIPAA-eligible environment with Amazon Quick | Artificial Intelligence
AWS published a guide on using Amazon Quick, its agentic AI workspace, to handle protected health information (PHI) under HIPAA. The post…
Write your own training loop — without managing GPUs — with Amazon SageMaker training session APIs | Artificial Intelligence
AWS describes new SageMaker AI training session APIs that let developers write custom model training loops in their own Python process while…
Announcing new ways to customize models with SageMaker AI, without managing GPUs | Artificial Intelligence
AWS announced general availability of training sessions as part of Amazon SageMaker AI serverless model customization. The feature lets developers write their…
How IskayPet unified corporate knowledge with Amazon Bedrock Managed Knowledge Bases | Artificial Intelligence
Per an AWS Machine Learning blog post, Iberian pet care company IskayPet built an internal knowledge assistant on Amazon Bedrock Managed Knowledge…
Amazon Polly delivers a major upgrade to its Generative engine | Artificial Intelligence
Amazon Web Services announced a generally available upgrade to the Generative engine of Amazon Polly, its managed text-to-speech service. The update adds…
NVIDIA Nsight Python Tutorial: How to Profile GPU Kernels - MarkTechPost
This MarkTechPost tutorial walks through using NVIDIA's open-source Nsight Python library to profile GPU kernels from within a Python process, run on…
A Coding Guide to Cactus Whistle: Does a 16.9 MB Speech Model Hold Up Against Whisper Base? Word Error Rate, Speed, Streaming, Timestamps, Keywords, and Speech-to-Tool Calls on a CPU - MarkTechPost
MarkTechPost published a hands-on tutorial benchmarking Cactus Compute's Whistle, a 16.9 MB speech-to-text model that runs on CPU, against OpenAI's Whisper tiny…
Perplexity and NVIDIA Researchers Introduce JOLT: a Joint Teacher-Student Reinforcement Learning Method - MarkTechPost
According to a MarkTechPost report summarizing a paper from Perplexity and NVIDIA researchers, JOLT (Joint On-policy Learning and Teaching) is a post-training…
IBM AI Releases Granite Speech 5.0 2B for Multilingual Transcription and Translation - MarkTechPost
IBM has released Granite Speech 5.0 2B, a 2-billion-parameter open speech-language model that converts speech to text in six languages and also…
A Developer’s Guide to OmniParseBench: Leaderboard Benchmarks and OCR Testing - MarkTechPost
This MarkTechPost tutorial walks developers through OmniParseBench, an open benchmark from Datalab for document parsers. The benchmark grades parser output with 16,288…
Cohere Releases Rerank 5 With Passage-Level Context Focus and Calibrated Scores - MarkTechPost
Cohere has released Rerank 5, a retrieval reranker that reorders search results before they reach an LLM. It ships in two tiers:…
Gradium Open-Sources Phonon, a 31M-Parameter On-Device TTS Model Under Apache 2.0 - MarkTechPost
Gradium, a Paris voice AI company spun out of Kyutai, has released Phonon, a family of small text-to-speech models intended to run…
Liquid AI Launches RIPL, a Real-Time Inference Engine for Sub-50 ms LFMs - MarkTechPost
Liquid AI has introduced RIPL (Real-time Inference Pipeline), a low-latency serving engine for its Liquid Foundation Models (LFMs) aimed at live request…
Dynamic Early Exit in Reasoning Models
This arXiv preprint (Chenxu Yang and eight co-authors, first submitted 22 April 2025, revised through v3 in September 2025) proposes a method…
SocioVerse: A World Model for Social Simulation Powered by LLM Agents and A Pool of 10 Million Real-World Users
This arXiv preprint, authored by Xinnong Zhang and 20 co-authors, introduces SocioVerse, an LLM-agent-driven world model intended for social simulation. The authors…
AgentA/B: Automated and Scalable Web A/BTesting with Interactive LLM Agents
This arXiv preprint (Lu et al., submitted April 2025, revised through v4 in March 2026) presents AgentA/B, a system that uses LLM-based…
QuestBench: Can LLMs ask the right question to acquire information in reasoning tasks?
Researchers Belinda Z. Li, Been Kim, and Zi Wang (arXiv:2503.22674, first submitted 28 March 2025, revised 24 October 2025, listed as a…
Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning
Researchers from Skywork introduced R1V2, a multimodal reasoning model presented as a successor to Skywork R1V. Per the paper's abstract, the training…
nvidia/OpenMath-Nemotron-14B-Kaggle · Hugging Face
NVIDIA's Hugging Face model card describes OpenMath-Nemotron-14B-Kaggle, a 14B-parameter model produced by finetuning Qwen2.5-14B on a subset of the OpenMathReasoning dataset. The…
nvidia/OpenMath-Nemotron-32B · Hugging Face
NVIDIA has released OpenMath-Nemotron-32B on Hugging Face, a 32.8B-parameter math reasoning model created by fine-tuning Qwen2.5-32B on NVIDIA's OpenMathReasoning dataset. The model…
A Self-Improving Coding Agent
Researchers Maxime Robeyns, Martin Szummer, and Laurence Aitchison describe an LLM agent system equipped with basic coding tools that edits its own…
GitHub - XiaomiMiMo/MiMo: MiMo: Unlocking the Reasoning Potential of Language Model – From Pretraining to Posttraining · GitHub
Xiaomi's LLM-Core team released the MiMo-7B model series on GitHub, a set of 7B-parameter language models trained from scratch with a focus…
Qwen/Qwen2.5-Omni-3B · Hugging Face
This is the Hugging Face model card for Qwen2.5-Omni-3B, a 3B-parameter end-to-end multimodal model from the Qwen team (Alibaba Group) that accepts…