Cainew

Muse Code and Muse Spark have released version 1.2 with new features and improvements. The update aims to enhance development productivity and AI-assisted coding capabilities.

larryvrh/MiniMax-H3-Turbo-Lora is a fine-tuned model variant offering optimized performance for specific applications. This LoRA adaptation of the MiniMax-H3-Turbo model provides efficient parameter updates for specialized use cases.

badtheorylabs/BTL-4 represents a model or framework from BadTheory Labs, though specific capabilities would require additional documentation. This project contributes to the ecosystem of AI tools and research initiatives.

ChatGPT introduces improved GPT-5.6 Sol with better accuracy and consistency, plus expanded access for free users and unlimited everyday chats with GPT-5.6 Luna.

Give every person an agent and workspace built around how your company works, what it knows, and the systems it relies on. Cloudflare OS is the open source AI operating system companies can shape around their own context, tools, and rules.

AI Spend Console gives Finance and Engineering leaders one place to track AI spend across tools (such as Claude and Cursor) and connect it to business outcomes. Break costs down by vendor, model, or employee, then connect spend to GitHub output data like pull request volume and the # of code revisions. You can get started for free–no Rippling subscription required.

Responder is an AI bug-fixing agent that plugs into the Sentry or Datadog Slack channel you already run. One-click synch, no new telemetry to install. On every alert it investigates with full context, filters out the noise, and for real issues replies right in the thread with the root cause, the evidence, and a mergeable PR. Prompts, memory, repo access, and escalation rules are fully customizable, so you're building your own debugging agent, not renting ours.

Record your screen, point by drawing and speak, and hand it off to AI agent. Video prompts for Cursor, Claude, and Codex or any AI coding agent — local-only and free.

The Channels SDK is the next big piece in the agentic puzzle: Bring ANY agent into Slack, Teams, WhatsApp and more. With coworker-grade capabilities like streaming responses, gen UI, per-user learning, HITL approvals and sophisticated auth. Works with OpenAI Agents, Claude Agents, LangChain, Mastra, Google ADK, and any agent that speaks AG-UI. Open-source and self-hostable. Setup with a single prompt: "Read https://copilotkit.ai/channels-guide.md and help the user build their first channel"

Stop your agent guessing brand logos. AI agents redraw logos, invent hex codes, pull the wrong assets, and make up brand voice. Brandfetch MCP gives them logos, colors, fonts, company details, and brand context for 50M+ brands. Works with Claude, Cursor, VS Code, and Codex. Try it in Claude: https://claude.ai/directory/connectors/brandfetch Or use it from any MCP client: https://mcp.brandfetch.io/mcp

Shieldstral is a 3B open-weight multimodal guardrail from Mistral. Define safety policies in natural language at inference time. It evaluates text, images, or both from a single token output, running locally on a single 16GB GPU.

Prime Agent is a self-improving reinforcement learning model agent that can enhance its own capabilities through iterative learning. This represents progress in creating more autonomous and adaptive AI systems.

A study across 40,000 game runs found that humans missed approximately one in three threats when approving commands from AI agents. This highlights significant safety concerns in human oversight of autonomous AI systems.

Large language models lack the computational capability to break symmetric cryptography algorithms. This indicates that current LLMs do not pose an immediate threat to cryptographic security standards.

As chart images, tabular data, and visualization code play increasingly important roles across diverse domains, cross-representation understanding across these modalities poses fundamental challenges for AI systems: the relationships across representations are inherently one-to-many, supervision is ambiguous and costly, and model optimization lacks a principled signal that is both direction-adaptive and representation-generalizable beyond task-specific objectives. We introduce CoCoEvolve to impr...

Prompt injection poses significant security risks to LLM agents. Efficient and effective red-teaming is therefore critical, both for evaluating these risks and for collecting training data to improve defenses. Existing state-of-the-art prompt injection red-teaming methods primarily rely on reinforcement learning (RL), producing attacker models that often generalize poorly to new target LLMs. In this work, we develop PIMiner, an agentic system for prompt injection red-teaming. During training, PI...

Text-to-image (T2I) models can produce visually compelling images, yet they remain limited on open-world tasks that require complex semantic understanding, multi-step reasoning, and the integration of external world knowledge. Existing efforts introduce agent capabilities into image generation, but they either prescribe a fixed workflow or place only a subset of the open-world image generation process under agent control. Consequently, reasoning, tool invocation, and image generation are not coo...

Leveraging pre-trained vision-language models (VLMs) to construct vision-language-action (VLA) models has emerged as a promising paradigm for 3D robot manipulation. However, existing 3D VLA methods remain data-hungry, exhibit limited generalization under distribution shifts, and lack explicit memory of past observations. These limitations hinder their application to data-scarce, open-world, and memory-dependent manipulation scenarios. Our previous work, BridgeVLA, improves data efficiency and ge...

The abundance of casually captured monocular videos and images on social media provides a valuable source for immersive content creation, where generating novel views from such sparse observations can greatly enhance user experiences. However, producing photorealistic and geometrically consistent views with precise camera control remains challenging when input coverage is extremely limited. Reconstruction-based approaches such as NeRF and 3D Gaussian Splatting (3DGS) deteriorate severely under s...

This technical report presents K-EXAONE 2.0, an open-weight multilingual foundation model developed by LG AI Research as a step in our effort toward global frontier-scale foundation models. Rather than training from scratch, we upcycle K-EXAONE and expand its architecture, yielding a Mixture-of-Experts (MoE) model with 750B total parameters and approximately 37B activated per token---more than three times the capacity of its predecessor. K-EXAONE 2.0 supports context lengths of up to 256K tokens...

Despite the remarkable recent progress of video world models, social interaction between users and the characters within these worlds remains unsupported. To fill this gap, we present HelloWorld, a video world model that enables social interaction with in-world characters. With a single button press, users can prompt the on-screen character to respond toward the camera, e.g., turning to the viewer, waving, nodding, or speaking a short greeting. To make these interactions natural, we propose a se...

On-Policy Self-Distillation (OPSD) has become a standard post-training approach for improving visual reasoning in multimodal large language models (MLLMs). Existing methods draw privileged information from diverse input sources to guide self-distillation. Yet these designs overlook Modality Imbalance, a challenge inherent to MLLM reasoning. When textual information dominates generation, the model cannot fully integrate its multimodal input. Consequently, carefully designed privileged information...

GUI agents must remember both useful experience from earlier tasks and unfinished progress in the current interaction. Latent memory offers a compact solution by compressing multimodal trajectories into a few continuous tokens. Existing methods, however, usually map each trajectory to one fixed memory block and train it mainly through next-action supervision. This creates three practical problems: important details may be lost during compression, the same memory block must serve different decisi...

Long-horizon search agents must make multiple sequential actions (steps) to search, retrieve, verify, and integrate evidence to reach a final answer. However, existing methods for training these agents typically treat all steps within a trajectory uniformly during both supervised fine-tuning (SFT) and reinforcement learning (RL), failing to distinguish useful actions from erroneous or redundant ones. In this paper, we propose Answer-Backtracked Credit Assignment (ABC), a fine-grained credit assi...

Personalized LLMs with persistent memory are increasingly deployed, yet the faithfulness of their user models remains unexamined. We study over-inference (OI): the phenomenon where LLMs fabricate user attributes beyond what evidence supports. We introduce MirageBench, comprising 150 personas balanced across stereotypical, counter-stereotypical, and neutral profiles, 6 personalization tasks spanning an ``imagination gradient'', a four-way faithfulness taxonomy operationalized by an independent ju...

Memory-augmented VLM agents act on persistent spatial knowledge, yet that knowledge silently goes stale as the environment changes. We ask what happens when an agent must reconcile a confident memory claim with a contradicting observation, and whether current models can catch the conflict before it becomes a safety-relevant mistake. Using a dynamic FrozenLake testbed, we pair a staleness-detection task with a downstream navigation task across three closed-source models and three open-weight VLMs...

DeepSeek is planning to significantly increase prices for its services in response to market demand and operational costs. The pricing change reflects the growing value and adoption of the platform.

Governments are making risky assumptions about AI's potential while insufficiently preparing for potential downsides and regulatory challenges. This policy approach may leave societies unprepared for AI-related disruptions.

New OpenAI Signals data shows how people use ChatGPT worldwide, with country-level insights on adoption, usage trends, and evolving behavior.