Cainew

Ternlight is a compact 7 MB embedding model capable of running directly in web browsers via WebAssembly, enabling on-device AI processing.

Badge gives an AI agent that collects peer reviews to build proof of work. For job seekers - The agent connects to your contacts, requests anonymous reviews from your past colleagues, and builds a verified, portable trust score. score. For hiring managers - Badge gives you a 30 second reference checks on candidates, collecting authentic feedback from their actual coworkers to replace unreliable, AI-generated resumes and LinkedIn Recommendations.

Your customers ask AI before they ask Google, and most brands have no idea whether they show up or lose to a competitor. Every other tool hands you a visibility score and stops there. Scribble does the whole loop: audit where you're invisible across every AI engine, create content that closes the gap, and amplify it through 50,000 creators who only get paid when AI cites them. A gap report won't win customers. Being the cited answer will.

Katalyst is the AI sales agent for teams on Salesforce. Hang up a call and it's already done: notes summarized, records created, fields updated, follow-up drafted, next step set. It runs 24/7, reading every call, email, and calendar, surfacing the right signals, prepping you for each meeting, and prompting you to act: follow up here, this one’s slipping. What's new: AI Resolution on every account, meeting recorder, hygiene scores, deal patterns. Built for enterprise sales teams.

Unlike AI tools that stop at interview + transcript, Mira is a full AI researcher — plans studies, recruits globally (100M+ panel, 120 countries), runs dynamic interviews with intelligent probing, and uniquely captures what participants say AND feel via real-time facial coding, voice emotion AI, and webcam eye tracking. Extracts themes, generates insights, and produces research reports automatically. 17 patents. 70+ languages. Trusted by Unilever, Nestlé and 150+ global brands. $25M Series B.

Ellis is an AI notetaker for in-person meetings. Record your meeting, get a clean transcript with each speaker identified, then ask anything — what was decided, what you missed, how it went. No laptop. No extra hardware. Just your iPhone (or Apple Watch).

@O is the ultimate AI coworker that lives natively in Slack. Tag @O like a colleague, to ask anything or delegate daily tasks in plain English. It connects to 1,000+ tools your business runs on, does work while you sleep, and shares memory and skills across your whole team right in Slack, on any model you choose, including your own. One-click install, and everyone is AI-enabled in under 5min, not just your power users. Zero friction, maximum adoption.

LongCat-2.0 is an MIT-licensed 1.6T-parameter MoE model with ~48B active parameters, 1M context, LongCat Sparse Attention, and post-training for coding and agentic workflows. It was trained on AI ASIC superpods and integrates with Claude Code, OpenClaw, and Hermes.

Zoho Tables is a no-code spreadsheet-database hybrid with AI brains to organise work, automate workflows, and collaborate in real time.

We’re announcing new capabilities in Managed Agents in Gemini API so developers can build reliable, production-ready agents.

Poly/ML provides a standard ML implementation framework, offering a consistent approach to building and deploying machine learning solutions.

This technique optimizes Retrieval-Augmented Generation (RAG) systems by dynamically pruning context to only include information actually needed for accurate answers.

We introduce the first multiplayer world model for highly dynamic environments governed by complex physical interactions. Whereas single-player world models treat the other agents as part of the environment, ours conditions on the action streams of multiple agents, learning to attribute changes in the scene to the correct player and to stay coherent under arbitrary combinations of their actions. We study this problem in the game of Rocket League, where players compete and cooperate under fast, t...

Pretraining scaling laws reveal that model capability improves predictably with data and compute. But learning from real world environments after deployment remains far less understood. Analyzing roughly 38,000 hours of agent interaction with the environment across 134 real world tasks, we find, to the best of our knowledge, the first evidence that overall performance during environment learning follows a log-sigmoid scaling law with remarkably high precision, reaching R^2 = 0.998. Across model ...

Unified models for robot manipulation aim to equip one policy with both the semantic priors of pretrained VLMs and the physical dynamics learned through future prediction. In practice, existing designs tend to erode the semantics of the pretrained backbone, suffer interference among heterogeneous objectives, and learn future prediction from scratch in pixel space, leaving the dynamics priors of pretrained video generators unexploited. We present InternVLA-A1.5, which builds the policy on a nativ...

Scaling pre-training, post-training, and test-time compute have become the central paradigms for improving the capabilities of LLMs. In this work, we identify verification, the ability to determine the correctness of a solution, as a new scaling axis. To unlock this and demonstrate its effectiveness, we introduce LLM-as-a-Verifier, a general-purpose verification framework that provides fine-grained feedback for agentic tasks without requiring additional training. Unlike standard LM judges that p...

3D reconstruction and generation are commonly tackled by separate paradigms: pixel-based regression for reconstruction, and latent diffusion for generation. Recent works attempt to unify them in latent space, but with notable drawbacks: the diffusion objective is defined on latent features rather than the underlying 3D representation, and both branches suffer from information loss introduced by latent encoding, while requiring a pretrained Variational Autoencoder (VAE) or Representation Autoenco...

Dense spatial perception is essential for physical intelligence, where visual systems are expected to recover structured, metric, and actionable representations from pixel observations. Modern visual foundation models tend to prioritize semantic invariance, often at the expense of detailed spatial understanding. In this work, we study vision pretraining through a boundary-centric lens, motivated by the premise that boundaries and shape discontinuities offer essential cues for perceiving geometri...

Audio intelligence involves understanding, reasoning about, and generating both audio and speech. In this work, we introduce Nemotron-Labs-Audex-30B-A3B (Audex), a unified audio-text LLM built on Nemotron-Cascade-2-30B-A3B, a strong text-only MoE LLM. Audex adopts a simple unified design with a single Transformer decoder: audio inputs are encoded and projected into the text embedding space, while text tokens and quantized audio output tokens are treated uniformly during generation. This architec...

Recent advances in video diffusion models have enabled either long single-view generation through temporal autoregression, or short multi-view synthesis through bidirectional attention. However, generating long, multi-view consistent videos of dynamic scenes remains unsolved. In this work, we present MV-Forcing, a framework that composes temporal and view-wise autoregression within a single diffusion model by introducing a 4D geometric bridge between sequentially generated views. Our key insight...

Predicting object dynamics (i.e., world modeling) is a fundamental challenge for robotic manipulation, and modeling deformable objects presents a particularly difficult case due to their high-dimensional state spaces and complex material properties. While current world models approach this through two distinct paradigms: learning the dynamics over the 2D pixel space or more explicit 3D geometric space. A systematic understanding of their relative strengths and limitations remains elusive due to ...

Multi-vector vision-language retrieval preserves fine-grained visual evidence through maximum-similarity late interaction, but dense image-side tokens make storage and scoring expensive. Existing token compression methods reduce this cost, yet they can remove or collapse object- and region-level evidence that future query tokens may need to select. We propose SaMer, an object-aware token merging framework that compresses image-side post-projector tokens into K representative centroids while pres...

Key-value (KV) cache growth is a major bottleneck in autoregressive decoding, as memory and bandwidth scale linearly with context length. Existing KV eviction methods often rely on static heuristics or proxy scores, which poorly track future token utility and cause brittle eviction as relevance shifts. To address this, we introduce KVpop, which learns a fixed-budget KV eviction policy by directly supervising the keep-or-drop decision. The scorer is trained against a novel future-attention target...

We present SynCity 3000, a framework for generating 3D scenes that are globally coherent while enabling fine-grained layout control. Building on the ability of current image-to-3D generators to produce complex 3D assets from a single image, we extend this capability to the scale of entire scenes by adapting the generator to be applicable as a convolutional operator. We achieve this by fine-tuning the model on scene-like data generated by a new synthetic data engine, which we propose to address t...

30papers.com curates 30 essential machine learning papers and presents them in a beginner-friendly format to make foundational ML knowledge more accessible.

Small AI models are gaining adoption in regions with unreliable network infrastructure, where their lower computational requirements offer practical advantages.

China's government is implementing restrictions on overseas access to its leading AI models to maintain domestic control over advanced technology distribution. This move reflects Beijing's efforts to regulate AI model availability beyond its borders.

The EU Parliament has passed the first reading of Chat Control legislation, advancing the controversial proposal toward potential implementation.

AI's return on investment timeline may be considerably longer outside the technology sector, where implementation challenges and industry-specific adaptations require extended development periods.

See how Australian Payments Plus uses ChatGPT Enterprise and Codex to move faster through payments complexity. AP+ saves time, improves quality, and keeps human judgment central.

GLM 5.2 and broader market trends suggest an impending margin collapse in AI as competition intensifies and model costs decrease.

Focusing on price per 1M tokens as a metric is misleading, as it doesn't account for model quality, latency, and other factors that affect actual AI application costs.