Cainew

Claude Opus 4.8 is a new version of Anthropic's flagship AI model with enhanced capabilities for complex reasoning and task execution.

LiquidAI/LFM2.5-8B-A1B is a compact language model designed for efficient inference while maintaining strong reasoning capabilities. This model represents advances in creating smaller, more practical language models for resource-constrained applications.

Arkon: Enterprise AI Knowledge Hub & MCP Server. Self-hosted knowledge base for teams to manage RAG contexts, access policies, and AI skills. Connect Claude and other LLMs via Model Context Protocol (MCP) for automated, secure organizational knowledge integration.

DIY OS bundle for the M5Stack Cardputer: Claude Buddy (BLE), Push-to-Claude (voice + chat with memory via a Cloudflare Worker), and a flash-and-go installer skill for Claude Code. Forked from moremas/build-with-claude.

Every other AI product is a tool that makes you more productive. A copilot. An assistant. A coworker. Something you use. Pancake makes your company autonomous. Agents with roles, goals, and a heartbeat working while you sleep. You set direction, approve the irreversible, the rest runs. Prepare yourself to be prompted by Pancake.

A hand-curated library of the best machine learning education — 590 docs (78 arXiv papers, 474 course lectures from Stanford/MIT/Karpathy/fast.ai, 38 explainer articles), normalized to Markdown with full provenance. A clean ML corpus/dataset for learning, RAG, and fine-tuning.

Most AI tools apply your colors to generic layouts and call it “on brand.” Pitch Agent builds from your template, design language, and image style. Generate slides from a prompt and file attachments, then refine them via chat. Agent lives inside Pitch, the workspace where teams collaborate on and deliver presentations.

Revolte is for engineering teams to turn intent into production-ready software faster, safer, and with more control. Its agents plan changes, generate code, run quality and security checks, create PRs, support deployment, monitor runtime behavior, and surface risks early. Engineers approve the important decisions. Revolte handles the delivery heavy lifting. Built for higher delivery throughput across SDLC, stronger governance, and more value shipped per engineer.

Buffer's API lets you publish and manage content across 10 social platforms through a single endpoint. Connect it to AI assistants, no-code automation tools, or build full custom integrations. Ships with an MCP server, pre-built automation templates, a CLI, and an interactive API explorer. Available on every Buffer plan, including Free.

Claude Code has introduced Dynamic Workflows, a new feature that enables more flexible and adaptive automation for complex coding tasks.

Customers can connect their own AI agents to Robinhood to help manage and automate trading and credit card purchases, with built-in safety controls and a real-time activity feed. Trade in a dedicated agentic account to stay in control of every trade your agent makes.

Memori launched its new agent-native memory infrastructure, enabling agents to create structured, long-term memory directly from agent trace — including execution paths, tool results, workflow steps, outcomes, and decision-making logic. This allows memory to also be generated from what an agent actually does. Benchmark results: 81.95% accuracy on LoCoMo using only 1,294 tokens per query, roughly 5% of full-context cost, saving users 95%+ on inference spend. 15K GitHub stars, 200000+ downloads

Generate personalised YouTube titles, descriptions, and thumbnails in minutes and update them based on video performance — built for creators exhausted by post-production work.

Drop your paperwork. Granite reads every document the moment you upload, files it correctly, and remembers it indefinitely. Find anything later by asking in plain English.

iPhones running iOS 26 implement automatic content detection that freezes FaceTime calls when nudity is identified, a privacy and safety measure introduced in 2025.

An open-source AI Racing Harness project provides tools and infrastructure for benchmarking and testing AI models in competitive racing scenarios.

Research shows significant disagreement among frontier large language models when fact-checking real-world claims, raising questions about their reliability for verification tasks.

Scientists have developed a Eureka machine that mimics natural exploration processes to discover solutions and research areas that current AI systems cannot independently identify.

World models for interactive video generation have largely focused on single-agent settings, where future observations are generated from a single control signal. However, many generated environments require multi-agent interaction: multiple players, robots, or embodied agents act simultaneously within a shared space. Scaling world models to such settings requires a principled multi-agent design: agents should remain independently controllable, permutation-symmetric, and support efficient infere...

Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as the standard paradigm for improving reasoning capability of large language models, while Multi-Token Prediction (MTP) has been a widely adopted module in pretraining. Combining them is a natural approach, yet current RL practices detach MTP gradients because joint training degrades the performance. We revisit this failure from an optimization perspective. We show that the per-step effect of MTP on the RL objective can be decomp...

Vision-language models with extended reasoning succeed on complex problems, but many real-world problems require external tools that internal reasoning alone often cannot resolve. Agentic reasoning therefore interleaves two behaviors with a structural asymmetry: thinking (the self-contained default) and tool use (a high-variance auxiliary acting). We refer to this asymmetry as the Thinking-Acting Gap. Under standard RL recipes like GRPO, the gap manifests as two diagnostic symptoms during traini...

Recent advances in online reinforcement learning (RL) for large language models (LLMs) have demonstrated promising performance in complex reasoning tasks. However, they often exhibit an imbalanced exploration-exploitation trade-off, resulting in unstable optimization and sub-optimal performance. We introduce IB-Score, a novel metric grounded in Information Bottleneck theory that evaluates policy's exploration-exploitation balance by quantifying the trade-off between step-level reasoning diversit...

Computer-use agents (CUAs) have recently made substantial progress, but deploying a separate large expert for each software domain remains expensive. Small open computer-use agents are more practical specialization targets, but they remain substantially weaker and exhibit uneven domain-specific failures. A straightforward remedy is to synthesize large-scale training data for the target domain, yet we find that this naive approach yields only marginal improvements. Building on this observation, w...

Existing emotional support conversation (ESC) systems mainly rely on end-to-end response generation or coarse strategy supervision, offering limited interpretability and little support for systematic skill improvement. We propose ESC-Skills, a skill-centric framework that discovers and self-evolves executable emotional support skills. We first model localized support interactions as Intervention Units (IUs), which capture state--action--outcome dynamics between seeker states, support interventio...

Are LLM-based search agents genuinely searching, or using the web to verify what they already know? We study this question on BrowseComp with three diagnostics. Our analysis reveals Intrinsic Knowledge Dependence (IKD): even with tool access, agents often rely on intrinsic knowledge -- information encoded in the model before retrieval -- rather than on external evidence. Agents answer up to 44.5% of BrowseComp questions without tools, generate more than half of their search queries from internal...

Reinforcement learning has become a central paradigm for advancing reasoning in large language models, yet most existing methods still depend on stronger teacher models or heavily curated difficult datasets, limiting scalable capability improvement. In this paper, we introduce DenoiseRL, a reinforcement learning framework that substitutes external supervision with recovery-oriented optimization over failures from weak models. Instead of relying on stronger supervision or carefully engineered dat...

Embodied Vision-Language Models (VLMs) have demonstrated impressive performance and generalization in robotics, particularly within Vision-Language-Action frameworks. However, a significant gap remains between the high-level semantic focus of standard text-guided pre-training paradigms and the low-level spatial and physical knowledge critical for execution in embodied environments. In this paper, we introduce GEM, a Generative-supervised Embodied vision-language Model designed to bridge this div...

Hybrid-reasoning large language models (LLMs) expose explicit controls over reasoning effort, allowing users or systems to trade off answer quality against inference cost. However, existing methods for adaptive thinking-mode selection are typically evaluated under different models, datasets, and implementation assumptions, making it difficult to compare their practical behavior. We introduce HRBench, a unified evaluation framework for studying thinking-mode switching in hybrid-reasoning LLMs. HR...

Despite the rapid progress of multimodal large language models in building Graphical User Interface (GUI) agents, their real-world task completion is fundamentally bottlenecked by a lack of world knowledge about GUI operations. Existing solutions typically rely on expensive multi-agent scaffolding or conventional post-training paradigms, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). However, post-training only allows agents to implicitly absorb world knowledge through act...

Visual outcomes are increasingly central to multimodal large language models, making reliable and fine-grained verification essential for scaling generalist foundation models. In this work, we investigate multimodal meta-verification, which leverages verifier-generated rationales rather than decision-only signals, and explore how to effectively incorporate meta-verification feedback into multimodal verifier training. We identify two key findings. First, symbolic verifier outputs (e.g., bounding ...

AI research agents can now generate research ideas, design experiments, run code, and draft papers, raising the possibility of large-scale AI-assisted scientific discovery. Many current agent frameworks explicitly encourage the generation of novel and high-impact ideas. Yet it remains unclear whether AI-assisted ideation broadens scientific exploration or mainly concentrates around existing work. We study AI research agents as scientific search systems. Using four AI research-agent frameworks an...

Anthropic has secured $65 billion in Series H funding, valuing the company at $965 billion and solidifying its position as one of the leading AI development companies.

Explore OpenAI’s Frontier Governance Framework and how our AI safety, security, and risk practices align with emerging EU and California regulations.

YouTube announced plans to automatically label videos created with AI-generated content to improve transparency and help viewers identify synthetic media.

Corporate America is facing substantial cost increases from enterprise AI implementations, as organizations grapple with licensing, infrastructure, and operational expenses.

Here are 12 of the biggest Google I/O 2026 keynote moments, including news about Gemini Omni, Gemini 3.5 Flash and more.