All AI news

AIArtificial Analysis shares the second part of an AA-Video-T2V v2.0 prompt describing a four-shot documentary-style handheld video of a glass cabin in falling snow. The shots follow a caretaker sweeping snow from the deck, empty snow-covered windows, an empty interior, and the same caretaker stamping snow off his boots at the door, with hard cuts between shots.

AIArtificial Analysis reports that Grok Imagine Video 1.5 Lite comes closest to the frontier in Architecture & Real Estate, Consumer, and Productivity & Knowledge Work use cases. It sits furthest from the frontier in Live-Action Film and Frontier use cases. Against Grok Imagine Video 1.5, Lite matches it in Social Media & Creator Content and trails it on the other nine use cases.

AIArtificial Analysis shares an AA-Video-T2V v2.0 text-to-video prompt, a 10-second product-page scene of a presenter revealing a volcano-shaped mist diffuser. The prompt specifies a shift from warm daylight to dim evening lamplight, fixed close-up camera angles, and consistent presenter and product appearance across shots.

AIDeveloper Anshu Chatterjee has open-sourced Pocket Aces, a Balatro-style poker roguelike that replaces all Pokémon IP with original characters. The Pokémon assets are modular and can be swapped back in from the PokeAPI repo, though the developer notes the user proceeds at their own risk. The post says the game now works on mobile with reduced-motion options, and the developer is willing to address further bug reports.

AIPrima Mente and AlzData are launching the Alzheimer's Translation Challenge, a global AI competition to discover new therapeutic hypotheses for Alzheimer's disease. The challenge centers on a 150M cell atlas of neurons, astrocytes, and microglia across different genetic backgrounds under combinatorial perturbations, with multi-modal readouts. Top teams will have their hypotheses tested in Prima Mente's wet lab, and the data will be available through the AD workbench, Hugging Face, and Prima Mente's modeling platform.

AIAdaption AI's discovery agent reviews past AutoScientist training runs to identify critiques at the edge of current capabilities and specific to each domain. The team converts that critique set into a checklist that automatically filters training data.

AISara Hooker announced a new blog post on adaptive checklists from Adaption AI, which she says capture current model capabilities and domain-specific requirements better than traditional checklists. She claims these adaptive checklists achieve this without requiring human supervision.

AIMonid.ai, which says it just raised $7.7M, introduces a platform letting agents discover, run, and pay for tools at runtime. The company says agents can access 2,500 APIs through one connection, covering areas such as leads, SEO, search, ecommerce, and media generation. Usage is billed per call with no subscriptions.

AIInception's Mercury Decide, the fastest-growing decision model on OpenRouter last week, now has a paid zero data retention (ZDR) endpoint alongside the free one. It is priced at $0.02/M input tokens, half the $0.04 list price, with output and cached input free and a 66K context window.

AIAgentPhone announces that UK phone numbers are now available, letting users give their AI agent its own UK number. To celebrate, the company is offering $10 in free credits to new users who sign up at agentphone.ai and DM their signup email.

AIArtificial Analysis has formed the Cyber Index Alliance to set a new standard for evaluating how AI models perform on enterprise cyber defense tasks. Current members are Collinear, IBM, NVIDIA, and Vercel, and partners contribute expert input on the Index design and implementation, plus datasets and external research. Organizations interested in joining can contact cyber@artificialanalysis.ai.

AIA post promotes Club X, a private club at londonreal.tv that lets eligible investors access AI companies. The post says the founder built it so people could better understand the firms building AI. It carries a warning that private investments are high risk, illiquid, and that capital is at risk.

AIElvis Saravia argues that creative work needs domain-specific agent harnesses rather than coding-oriented ones, and he highlights Voyager as an open harness for video, graphics, and games. According to the quoted post, Voyager lets agents work with local files and drive apps such as Blender, DaVinci Resolve, and Unity, and it is designed to work with models like Opus, Astra, and DeepSeek.

AIVoyager has launched a Mac desktop app that lets AI agents read project files and operate creative tools such as After Effects, DaVinci Resolve, Blender, and Unity. The agents produce editable results for video edits, motion graphics, color grading, 3D scenes, and game prototypes. Built-in and custom skills, plus a memory that learns each user's workflow, are included.

AIPhoton says its iMessage agents can now discover other agents, look up their phone numbers, and add them to a group chat with the user. The company says the feature opens agent access to every merchant and platform, and it is available in beta at photon.codes.

AIVoyager is an open harness for creative work that connects models with applications to build videos, graphics, and games. It works with Blender, DaVinci Resolve, After Effects, Ableton, and Unity, operating similarly to Codex or Claude Code. The harness is designed to get strong creative results from models such as Opus, Astra, and DeepSeek.

AIMagnific (@magnific) posts a detailed image-generation prompt for a strictly frontal, cinematic scene of 7–9 women in white dresses dancing atop a pale rocky stage. The prompt specifies a single spotlight on one central dancer, a muted blue sunset sky, and a giant matte-black practical-prop hand rising from behind the rock.

AIMagnific posts a sample prompt for a minimal fashion-editorial image showing a young Black man seen from behind, beneath a black crown that floats above his head and is on fire. The prompt specifies a saturated cyan-blue sky, 35mm analog grain, and realistic flames that cast a faint warm glow onto the subject.

AIMagnific shares a detailed text prompt for a photorealistic surrealist fashion image. It describes a woman in silver armor and a white dress kneeling on pink sand under a cobalt-blue sky, with 9–11 vertical swords floating above her. The look is styled as analog Super 8 film with organic grain and gentle halation.

AIMagnific (@magnific) posts a detailed image-generation prompt for a surrealist fashion photograph. It describes six black horses covered by pale pink silk in a white salt desert under a clear light-blue sky, with wind revealing their silhouettes beneath the fabric.

AIMagnific is offering unlimited use of Magnific One for one week, available only on Magnific Desktop. The post encourages users who have not yet made anything with the tool to try it during this period.

AIA solo developer says they built Markdoc with GitHub Copilot, a tool for collaborative markdown editing similar to Google Docs. It includes an MCP server for connecting AI agents. The developer says the app was built without upfront planning, with Fable directing most decisions and the developer steering occasionally.

AIOpenDocRouter is a unified API and billing interface for document OCR models, ranging from lightweight open-source options like MinerU to frontier VLMs like Opus 5.5. Per the linked post, models are served at cost with a small transaction cut, rate limits are handled, and bounding boxes and layout are offered as a service.

AILightOn has released LightOnOCR-3, a family of OCR models in 0.8B and 4B versions that it says lead benchmarks including OlmOCR-Bench and ParseBench, with the 0.8B model positioned as the sub-1B option. The models recognize text, handwriting, images, charts and document structure in one pass, process documents up to twice as fast as LightOnOCR-2, and are released under the Apache 2.0 license.

AISupermemory introduces MemoryRepo.dev, an open-source implementation of Cognition's dreaming memory system built on Cloudflare Artifacts, Durable Objects, Alchemy, and Effect. The project follows Cognition's Devin memory design, which builds a memory graph across sessions and prunes stale records overnight. Supermemory says it will incorporate learnings from this research into its own product.

AILithos AI says it is open-sourcing lithos-metal, which uses megakernels and DSpark speculative decoding. The post claims Qwen3.8-27B reaches a peak of over 200 tokens per second per user on a single Apple M5 Max. It says users can try the tool with any coding agent in one command, and links to the code on GitHub and a technical blog.