Updated
Updated
Items with an AI score under 20 are hidden. Show low-relevance items
Perplexity@perplexity_aiAI score27AIPerplexity has published both pplx-embed-v2 models on Hugging Face, with the collection available at the linked URL. The post provides no further details on model sizes, benchmarks, or capabilities.
Perplexity@perplexity_aiAI score32AIPerplexity reports its system reached 92.4% accuracy on MADQA, a benchmark of 500 questions over 800 PDFs. The post describes this as the top result among retrievers on that benchmark.
Perplexity@perplexity_aiAI score34AIIn agentic search, a GPT-OSS-120B agent using the 9B model answers 64.0% of BrowseComp+ questions correctly, 4.9 points ahead of the next ColBERT model. It also uses fewer searches than any baseline.
Perplexity@perplexity_aiAI score22AIOn the Q2D-Web benchmark of about 70,000 production-derived search queries over 190M web documents, Perplexity's 9B and 0.6B models reach 74.8% and 73.6% Recall@1000. Both outperform nemotron-embed-8b, which scores 69.3%.
Andrew Curran@AndrewCurran_AI score42AIAndrew Curran notes OpenAI's 722 published mathematical results show striking under-representation of cryptographic breakthroughs, and he says he has personally witnessed US government censorship of academic quantum cryptanalysis results. He calls backroom government interventionism his base case and says rumors suggest yesterday's OpenAI math release was only the first of three batches.
Perplexity@perplexity_aiAI score41AIPerplexity's 0.6B and 9B models are both distilled token by token from one 18B teacher, so they share a single embedding space. A corpus indexed with the 9B model can be searched using 0.6B queries, raising ViDoRe v3 from 62.3% to 63.5% with no added query cost.
Perplexity@perplexity_aiAI score36AIPerplexity's models embed images and rendered pages directly, so PDFs, slides, and scans can be searched without OCR. This preserves tables, figures, and layout that text extraction typically drops.
Perplexity@perplexity_aiAI score36AIPerplexity's pplx-embed-v2-late retains a 128-dimensional vector for each token rather than compressing a document into one vector. It scores matches with MaxSim, pairing each query token with its closest document token, which the post presents as preserving detail in long or visually dense pages.
Microsoft Research@MSFTResearchAI score24AIMicrosoft Research introduced Agent Lightning, a tool that connects existing AI agents to reinforcement learning training. It aims to make agents easier to improve without rebuilding them, since their tools, context, and decision-making are typically managed by complex frameworks.
Perplexity@perplexity_aiAI score45AIPerplexity has released pplx-embed-v2-late, two late-interaction embedding models that retrieve text, images, and pages within a shared embedding space for cross-model querying. The company says both models achieve frontier performance and are publicly available on Hugging Face.
Jerry Liu@jerryjliu0AI score47AILlamaIndex has launched OpenDocRouter, a unified API that transcribes documents to markdown across frontier and open-weight models at cost plus a small transaction fee. The service handles rate limits, adds bounding boxes and layout as a service, and will benchmark new OCR models on ParseBench before adding them.
LlamaIndex 🦙@llama_indexAI score47AILlamaIndex announced OpenDocRouter, a single API that routes document parsing requests to any of 10 frontier and open-source models at launch, including Claude Opus 5.5, Gemini 3.8 Flash, GPT-6 Luna, MinerU2.5-Pro, and PaddleOCR-VL-1.6. Users can switch models in one line with the same request and markdown output, and each model is scored on ParseBench for quality and cost. Pricing is per-token, failed pages are not charged, and the service costs $0.86 to $48.82 per 1,000 pages depending on the model.
AIOpenAI's Codex 0.161.0 release makes GPT-6.1 Sol the default model in the bundled and Amazon Bedrock catalogs. It adds Amazon Bedrock support for multi-agent V2 and Ultra reasoning on compatible models, plus opt-in Daybreak routing enabled through --enable cli_daybreak or features.cli_daybreak=true.
AIMicrosoft Research Asia has open-sourced Agent Lightning v1.0, a roughly 3,500-line agentic RL framework that trains the same agent harness used in deployment. In an end-to-end coding agent pipeline, Qwen3.5-9B rose from 41.8% to 56.4% Pass@1 on SWE-bench Verified using about 6,000 training samples. The framework runs agents as standard Kubernetes jobs without paid commercial sandbox services.
Why it matters: The source shows how training with the deployed agent harness avoids rebuilding agents, and reports concrete SWE-bench Verified gains from about 6,000 samples.
AINVIDIA describes using digital twins and AI agents to validate changes to AI factory infrastructure, which combines GPUs, CPUs, switches, DPUs, and SuperNICs with schedulers, orchestration services, security controls, and a fast-changing software stack. The source frames the challenge as confirming that hardware, software, and policies work together for target workloads before deployment. The available excerpt does not give further detail on specific tools or results.
Ai2@allen_aiAI score36AIAi2 is making new Stage 1 checkpoints available, with the byte-level components already trained and the original model's weights left unchanged. Researchers can use them to test new architectures and train the full system without repeating that initial stage.
Ai2@allen_aiAI score20AIAi2 introduces Bolmo, a model that operates on bytes rather than tokens, which the team hopes will lead to larger models adapting how they represent information across languages and domains. Because bytes can also represent images and audio, the approach could eventually extend beyond text.
Ai2@allen_aiAI score39AIAi2 converted Qwen 3 8B and Llama 3 8B into byte-level models named Bwen 8B and Blama 8B. Both come close to matching their source models' performance in Ai2's evaluations, and Bwen 8B outperforms Bolmo 7B across its aggregate evaluation suite.
Ai2@allen_aiAI score36AIAi2 describes a recipe that adapts existing models to process raw bytes through a brief additional training run. The approach keeps the model's core intact and adds components that group bytes into variable-length patches for processing, then expand outputs back to byte-level predictions of the next byte.
Ai2@allen_aiAI score38AIAi2's paper on retrofitting language models to operate over bytes, the approach behind Bolmo, has been accepted to Nature. The team is also releasing new checkpoints that extend the method from Olmo to Qwen and Llama.
Demis Hassabis@demishassabisAI score34AIIsomorphic Labs has joined the Virtual Biology Initiative (VBI) as a founding member, helping build an open resource for global researchers to accelerate understanding and treatment of disease. Demis Hassabis, owner of the account and Google/Gemini-affiliated, called applying AI to medicine and human health the most important thing AI can be used for and welcomed the partnership with CZI and VBI.
AIAWS Machine Learning Blog argues that the traditional hours-saved ROI model, built for rule-based RPA, misses most of the value of agentic automation. It proposes an Agentic Value Model covering time savings, exception handling, decision quality, and change resilience, with value counted only when tied to a defined P&L mechanism and owner.
AIQlik built Qlik Answers, a natural-language assistant that returns sourced answers from knowledge bases, analytics apps, glossaries, and documents, using Amazon Bedrock for model access. The system routes each question through specialist agents and retrieval on Amazon OpenSearch Service, with Amazon Bedrock Guardrails applied to every request and response. Qlik serves more than 40,000 customers across regions, using Amazon SageMaker AI as an in-Region fallback when models are not yet available on Bedrock.
AIThe AWS Machine Learning Blog describes an automated remediation workflow that acts on AWS DevOps Agent investigation results. Amazon EventBridge triggers a Lambda durable function that uses Amazon Bedrock to propose fixes from an allowlist of tools, running read-only actions autonomously and pausing for human approval before infrastructure changes. The post demonstrates the flow with a Lambda function whose 3-second timeout is raised to 30 seconds after a single approval.
Lucas Beyer (bl16)@giffmanaAI score36AILucas Beyer praises Reality Check, a new leaderboard for benchmarking VLA and related robot manipulation models. Half of its tasks are fully open, while the other half are held out to detect benchmaxxing by future model versions. The companion post from Nicolas Keller describes the launch as the first public robot manipulation benchmark, built on 14,400 real-world rollouts across four models.
AIGitHub has made local sandboxing for GitHub Copilot generally available in GitHub Copilot CLI, the GitHub Copilot app, and VS Code sessions using Agent Host. Sandboxes restrict the filesystem, network, and credentials that Copilot-initiated tools and commands can access, based on developer or organization policies. The feature is powered by Microsoft eXecution Container (MXC), supports Windows, macOS, and Linux, and is included at no additional cost.
AIGitHub Copilot CLI version 1.0.94-0 lets users run /model to discover supported models from a running local Ollama instance alongside configured and GitHub Copilot cloud models. Discovered models are not added automatically; users choose one, review its provider and endpoint, then confirm Add and use for this session or Add without switching, and models must support tool calling and streaming. Choosing a local model does not enable offline mode or disable GitHub telemetry, and COPILOT_OFFLINE=true remains a separate explicit setting.
AINVIDIA's cuOpt decision optimization software now targets problems with 100 million variables using mPDLP. The source says cuOpt already delivers more than 10x speedups over CPU solvers, while detailing little else about mPDLP's design or availability.
AIAWS ran a six-week program pairing non-engineering professionals with mentors and tools like Amazon Bedrock AgentCore and the Strands Agents SDK to build working AI prototypes. Four participants with no engineering background built WealthWise, a multi-agent financial advisory tool with five agents on Amazon Nova models, which won first place. The article says participants who completed the phased program retained three times more practical skills than those in two-day intensive formats.
🚨 AI News | TestingCatalog@testingcatalogAI score20AIGoogle has rolled out a new Playground experiment in the US, letting paid-plan users generate content while all users can play games in its gallery. The post does not specify what kind of content can be generated or which plans qualify as paid.
Andrew Curran@AndrewCurran_AI score27AINous Research has raised $90 million at a $1.5 billion valuation, according to Andrew Curran's post. He praises the team and says Hermes has been a major success for the company.
AIThe Department of Defense's Chief Digital and Artificial Intelligence Office runs the Tradewinds program, which grants "post-competitive" status to AI vendors based on videos of five minutes or less. That status can let government buyers use other transaction agreements and sometimes make awards in less than a week. OpenAI, Anthropic, and Google are listed as participants, though none commented.
indigo@indigoxAI score34AIThe poster says they already use Grok Bot as a model router, citing last weekend's personal agent livestream. In the demo, Gemini produced an infographic inside Grok Bot, and Claude Opus then checked the content. This follows Elon Musk's announcement that Grok Bot will use the best backend model for each task, including Claude Opus 5.5, MidJourney, and Suno.
Waymo@WaymoAI score27AIWaymo has introduced a first-of-its-kind framework for autonomous vehicle incident-management exercises, ranging from tabletop scenarios to full-scale drills. Adapted from emergency management best practices, it is designed to help AV developers, operational partners, and first responders test plans and strengthen coordination together.
🚨 AI News | TestingCatalog@testingcatalogAI score41AIGoogle released the Google AI Edge Foresight app for macOS, powered by Gemma 4 and EmbeddingGemma 2. The app can connect to Google Drive to build a knowledge graph and, when transcription is active, uses local Gemma 4 E4B or Gemma 4 12B models to transcribe voice notes into new documents. EmbeddingGemma 2 is an open-weight, Apache 2 licensed 740M-parameter multimodal embedding model with an 8K context window.
Stanford HAI@StanfordHAIAI score31AIStanford HAI has released a new report examining partnerships in which funders have invested hundreds of millions of dollars to help nonprofits deploy AI for social good with frontier AI labs. Drawing on interviews, the report investigates who benefits from these partnerships and what drives their success or failure.
OpenRouter@OpenRouterAI score42AIEleven v4 is their most expressive voice model, with audio tags like [whispering]. v4 Turbo keeps that quality for real-time agents, Flash v2.5 is built for ultra-low latency, and Scribe v2 Medical is tuned for clinical conversations.
OpenRouter@OpenRouterAI score34AIElevenLabs is now available on OpenRouter, offering 9 Text to Speech models supporting up to 90+ languages and 2 Speech to Text models with speaker labels and word-level timestamps. All ElevenLabs models are 50% off on OpenRouter through October 19, 8am PT.
LangChain@LangChainAI score22AILangChain announces LangSmith Engine v2, which red teams agents, detects more issue types, and automatically tests proposed fixes. The update was presented by Ben Tannyhill at the Interrupt NYC session.
elvis@omarsar0AI score44AINVIDIA's VERA turns benchmark trajectories into over 9,000 restartable sandboxes with rubric scoring and updates both model weights and the agent harness together. A harness edit is kept only if it adds at least 5 points on the development set, and a checkpoint is rejected if its score drops more than 20%. At 27B, the co-evolved agent scores 71.6 on AutoCoWorkBench, above Claude Opus 4.8, and the environment corpus is open-sourced.
