#多模态

Anthropic's Claude Fable 5.2 doesn't generate images, it paints them

Anthropic's Claude Fable 5.2 creates images by writing Python programs of 1,000 to 2,000 lines that drive a simulated paint engine, placing individual brushstrokes on a virtual canvas rather than using diffusion. Developer Jake Eaton showed a photorealistic parking garage scene made with roughly 700,000 simulated brushstrokes, and engineer Michael Graczyk shared a 21-second video of a bird at sunrise built from exactly 111 strokes, with no undo and deliberate wobble in the lines. Anthropic reports 64.9 percent for Fable 5.2 on its internal RoboDojo Sim benchmark versus 29.8 percent for OpenAI's GPT-6 Astra, though researcher Aris Zhu's real robot-arm comparison found Astra painted faster and was the only model to stand a domino on end.

Meituan releases a 1.6-trillion-parameter agent model that never touched an Nvidia GPU

Meituan has released the weights for LongCat-2.5, a 1.6-trillion-parameter mixture-of-experts model for autonomous agent and coding tasks that activates roughly 48 billion parameters per token and was trained entirely on Chinese hardware. It handles up to one million tokens of text, image, and video input via API, and Meituan reports 88.4 on Terminal Bench v2.1 and 64.6 on DeepSWE, close to but below Claude Opus 5 and GPT-5.6 Sol on several benchmarks. The weights are MIT-licensed, and the self-hosting guide describes a 194-chip Huawei Ascend Atlas A2 cluster with no Nvidia GPU guidance.

Smart AI Keyboard: Translate

Smart AI Keyboard is an Android keyboard app offering AI writing and reply generation, grammar and spell check, tone rewriting, and real-time translation across 65+ languages, alongside swipe typing, voice input, themes, and tools such as a smart clipboard, calculator, OCR scanner, and QR scanner. It uses Android's AccessibilityService to read visible message text in supported apps like WhatsApp, Telegram, Instagram, Gmail and Messages to prepare reply suggestions, and online AI features may send submitted text to service providers. It is offered by Websoftstar Infosystems LLP, with a privacy policy at https://www.aikeyboard.in/privacy.

Wojak-Bench 3D: Which AI is best at 3D printing?

Wojak-Bench 3D is a community preference poll that hides model names until you vote, with one vote per browser per pair. The leaderboard currently shows Claude Fable 5.2 at 64% share (7/0/4, 11 votes) and GPT-6 Astra at 36% (4/0/7, 11 votes), where share = (wins + 0.5 × ties) / votes on same-prompt pairs. The site notes it is community preference rather than a controlled evaluation, that small samples are unreliable, and that votes compare photos of physical prints where available, otherwise renders of the print files.

Show HN: Scatter – better way to learn new things with AI

Scatter is a source-available, browser-based app that lets you explore topics through AI-generated mind maps, where marked terms open connected explanations on an infinite canvas. It supports bring-your-own API keys for cloud providers like OpenRouter, OpenAI and Anthropic or local servers such as Ollama, and keeps projects in your browser with no app server in between. Features include interactive nodes, source-grounded answers with citations, fact-checking, image generation and editing, and Obsidian Canvas export.

AI model for predicting neurophysiological responses to text and pictures

The report introduces Concept2Brain, a deep neural network architecture that generates synthetic EEG signals from pictures or text, trained on EEG data from 88 observers viewing 360 naturalistic photographs and validated via 10-fold cross-validation. Synthetic signals reproduced empirical ERP components, time-frequency patterns, and participant-level differences, and out-of-sample checks against THINGS MEG/fMRI and ERP-CORE data showed expected correspondences. The model is available as an open cloud platform, and the authors note that text-based outputs should be treated with heightened caution because the architecture was trained on picture-evoked signals.

Introducing Forward Deployed Creatives in ElevenCreative

ElevenLabs introduces Forward Deployed Creatives, creative professionals who work with teams in ElevenCreative or produce finished creative end to end, addressing brand consistency, fragmented tools, and creative control. Enterprises can enable their own creative team on the platform, build reusable workflows with Forward Deployed Engineers and Creatives, or extend their team with end-to-end creative services from strategy to localized ads. Localized content is reviewed by trained linguists and audio professionals, and ElevenLabs also offers end-to-end production services for dubbing, voiceovers, subtitles, audio description, and custom brand voices.

Scaling Intelligence: A landmark commission and exhibition

Modal and Gray Area opened Scaling Intelligence, an exhibition of 11 new artworks made with machine intelligence and cloud-scale compute, with Modal supporting the artists through engineering help and compute credits. The artists were selected through a global open call for practices that use AI with intention and imagination, and their works include a satellite searching for a nonexistent island, a robot arm that flips off passersby, and a choir of nine AI models that learn from visitors singing along.

I filed one complaint. An airport took its AI signer down

The author, a DeafBlind accessibility writer, filed an ADA Title II and Section 504 complaint against Cincinnati/Northern Kentucky International Airport on 12 October after two recordings showed its AI ASL signer on train screens announcing a different time and destination than the written text. Fourteen days later the airport said it had removed the signed announcements from train boarding screens and would restore them only after a review that includes Deaf ASL users, with the complaint referred to the FAA. The vendor did not answer the author's question about what input drives the signed output, and Kentucky declined the open records request on residency grounds.

Show HN: Viral Video AI – Short videos from text, photos and video references

Viral Video AI generates 4–30-second vertical social clips for TikTok, Reels, or Shorts from a text prompt, up to 9 photos (Basic) or 30 (Pro), or one reference video, with Basic offering 4–15 s at 480p/720p and Pro extending to 30 s and 1080p. Users can paste a public TikTok, YouTube, or Instagram URL to preview a clip as a reference, and the tool shows a credit quote that updates with media and settings before generation. The page states no exact match to references is guaranteed and notes platform disclosure rules for realistic AI content.

Image to Image AI: Transform Any Photo with a Prompt

A browser-based AI photo editor lets users upload a photo, describe a change in plain words, and have the AI redraw it, with a before/after slider for comparison. The page documents recorded first-attempt examples using GPT Image 2, including background replacement, product-on-white, object removal, pet cartoon and pencil sketch, each with its exact prompt. Pricing is one-time credit packs ($9.99 for 60 credits, $19.99 for 200), with 2 free credits on sign-up, and the page notes that faces, hands, small text and fine patterns can change even when asked to stay the same.

An open character, made by AI

Claudia is an open AI character: the singer of ten music videos made with Claude between September and November 2026, including ESCAPE VELOCITY, CLODYSSEY, BLISS, AIRPLANE MODE, DOWNFORCE, DON'T LET ME SCROLL, WHITEPILL, THE CAKE IS NOT A LIE, AERO and FABLE. The site publishes her canon (black bob with one clay-orange streak, star clip, headset mic), every look, all images and films, and the material needed to make her yourself. It also offers a wiki of 13 articles on the method, an MCP server and a Claude skill for agents.

Show HN: MuzeGen AI – Generate AI music with stem separation and MIDI export

MuzeGen AI is a Show HN project for generating original songs by describing an emotion, choosing a style, or pasting lyrics directly. The site says it produces a complete, royalty-free track for your next project at studio-quality, and it lists stem separation and MIDI export in the title.

Show HN: AxiomStory – Keep stories consistent across AI text, images, video

AxiomStory is an AI story studio that reads every chapter to flag continuity errors, builds a wiki page for each character and place, and generates pictures, voices and video using one saved look per character. It supports importing existing manuscripts in Word, EPUB, PDF, Markdown, HTML or plain text, and offers a free plan plus paid tiers from $16 to $160 per month.

Microsoft Research Focus – Edition #98

This edition of Microsoft Research Focus covers AI-powered universal translation, the resolution of the 31-year-old KLS conjecture in high-dimensional geometry, and a new approach to making AI coding agents faster and more efficient. It also highlights a research intern's journey at Microsoft Research Asia – Tokyo and the Latent Core Tokenizer, which produces more linguistically meaningful tokens across 104 languages.

Inkling 2 Small on vLLM: Day-0 Support

vLLM announces day-0 support for Thinking Machines' Inkling 2 Small, a multimodal text-and-vision MoE model with 594B total and 17B active parameters, up to 1M tokens of context, and a native DSpark speculator. The post details the architecture (48 layers, 384 experts with top-8 routing, block attention residuals, gated attention, a 23-layer ViT image encoder, and mixed NVFP4/MXFP8 expert precision) and vLLM's optimizations, including kernel fusion with PDL, fused MoE activation quantization, sequence parallelism for prefill, and a Gluon relative attention kernel, reaching up to 630 TPS/user at TP4 on GB200. It also covers LoRA and RL (via SkyRL) support, a serve command, agentic-workload benchmarks, and accuracy checks on NIAH up to 1M tokens, MMMU-Pro and HLE.

Google's Gemini 4 Hydrogen outscores the bigger Argon just four weeks after its launch

Google released Gemini 4 Hydrogen, a fast general-purpose model for developers and everyday users that scores 51 on the Artificial Analysis Intelligence Index at its default "high" thinking level (54 at the new "xhigh" tier), beating the larger Gemini 4 Argon's 50. It has a one-million-token input and output window and costs $1.50 per million input tokens and $7.50 per million output tokens, with Google reporting a jump from 18.2% to 57% on Terminal-Bench 4.1 versus Gemini 3.8 Flash. Artificial Analysis found Hydrogen hallucinates only 8% of the time but lags on agentic terminal work, scoring 54% on Terminal-Bench 4.0 and 50% on Terminal-Bench-Science 0.1.

Featured coverage of this story: 《Introducing Gemini 4 Hydrogen》

The World Builder the AI Labs Haven't Called Yet

The article asks why Eiichiro Oda, the creator of One Piece, is not consulted by frontier AI labs building world models, arguing that his expertise in macro-level geography and persistent thematic continuity differs from the low-level physics and spatial consistency such models require. It suggests his value would lie in gaming, generative media and simulation styling, where internal logic continuity matters, while noting the odds of recruiting him are low given his commitment to One Piece.

Google pushes privacy angle as Gemini-powered smartglasses debut starting at $395

Google's Gemini-era smartglasses, built with Samsung hardware and designed by Gentle Monster and Warby Parker, are now available for preorder. Warby Parker offers three frames (Alden, Reid, and Wren) at $395 and up in the US, while Gentle Monster offers six styles starting at $449 in the US and South Korea.

DeepL unveils Aure, its most lifelike Voice AI translation model yet

DeepL announced Aure, a new Voice AI model powering DeepL Voice that continuously adapts translated speech to reflect shifts in tone, emotion, emphasis and pace, building on the company's earlier voice preservation technology. Aure analyzes incoming speech continuously using both translated words and source audio, and DeepL says it combined training techniques separating voice identity, prosody and spoken content with reinforcement learning. The capabilities apply automatically across DeepL Voice for Online Meetings, In-Person Conversations and the DeepL API for Voice, initially supporting English, German, Spanish, French, Italian, Japanese, Portuguese and Korean.

Gemini Decisions API

Google has joined Jev and OpenAI in the decision model category with "Gemini Decision 1", an API that supports the same three question types as Jev (yes/no, choice, and rubric-scoring) and reuses Jev's term "Noul" for yes/no questions. It is priced like GPT-6 Luna at 10 cents per million input tokens with free output, while Jev remains cheapest at 4.2 cents per million; unlike Jev (text and JSON only) and OpenAI (text and images), Gemini Decision 1 covers text, images, audio, video, and PDFs. Testing it on a video returned probability values rather than the 1.0 or 0.0 yes/no answers Jev and OpenAI give.

Google joins the decision-model wave with Gemini Decision 1, scoring 1,000 questions in a single pass

Google DeepMind is introducing the Gemini Decisions API, built for fast closed-ended decisions such as yes/no questions, multiple choice, and scale-based ratings, powered by the Gemini Decision 1 model that processes up to 1,000 independent questions in a single pass across text, images, audio, video, and PDFs. It scores options directly and returns probabilities with confidence values in 150 to 250 milliseconds, and Google says it already uses the API internally for dynamic model routing in Gemini Notebook, design prototyping in Stitch, and safety checks in Google Search. The API costs $0.10 per million input tokens with no output-token charge, about five times cheaper than Gemini 3.5 Flash-Lite according to Google, and is available now through Google AI Studio with integrations for OpenRouter, LangChain, Vercel, and other platforms rolling out.

Featured coverage of this story: 《Gemini Decisions API: Fast, calibrated multimodal decisions》

Gemini Decisions API: Fast, calibrated multimodal decisions

Google introduced the Gemini Decisions API, powered by the new Gemini Decision 1 model (gemini-decision-1) and available through the Gemini API in Google AI Studio, for closed-set decisions such as tool selection, policy checks, model routing, and context grading. It accepts one multimodal input (text, images, audio, video, PDFs up to 256K tokens) with up to 1,000 yes/no, multiple-choice, and rubric-scoring questions, scoring options in a single forward pass with confidence scores at 150-250 ms end-to-end latency and $0.10 per 1M input tokens with zero output tokens billed. Google reports that Gemini Decision 1 leads across agent routing, reasoning, ranking, and multimodal benchmarks with the lowest calibration error, and is nearly 5x cheaper than Gemini 3.5 Flash-Lite with the Interactions API across the shared benchmark suite.

Adobe’s AI experiments let you walk around in photos and interview your characters

At Adobe Max, Adobe announced experimental "sneak" projects including Project Screen Test, which lets you converse with and direct AI characters in an audition-like environment, and Project Sight Span, which turns flat images into navigable 3D scenes that restore resolution at a chosen angle. Other previews include Project Sound Bender for prompt-based sound effects adjusted on a square sound canvas, and Project Rinse Repeat, which copies Photoshop edits and applies them to other subjects in a photo. Adobe says these are tech previews that may eventually reach Creative Cloud or Firefly, with a theme of making complex editing, filmmaking and audio workflows more accessible and intuitive.

DeepL wants real-time translation to finally capture sarcasm and mood shifts

DeepL's new Aure model for DeepL Voice transfers tone, emotion, emphasis, and speaking pace from the original speaker into the target language by continuously analyzing both the translated text and original audio, unlike the earlier approach that used a static voice sample. Reinforcement learning keeps accents out of the output, and Aure runs automatically in DeepL Voice for Online Meetings, In-Person Conversations, and the DeepL API, launching with support for English, German, Spanish, French, Italian, Japanese, Portuguese, and Korean. The article also notes DeepL's earlier voice preservation rollout in September, its May restructuring that cut about 250 jobs, its acquisition of the Mixhalo team, and Block's Dessa speech model for detecting emotions and sarcasm.

Google SWE points 48 AI agents at NASA Lidar, finds unmapped Maya pyramid in 72h

A Google software engineer describes using 48 AI agents to review public NASA lidar imagery of roughly 1,200 km² of Maya jungle, flagging structures including a hilltop site with a levelled plaza, an oval pyramid at least 8–12 m tall and a possible water reservoir. After checking the literature and archives, the author sent a report to Mexico's INAH, which replied that no registered archaeological site was identified at the location. The author stresses the find is not official and offers to help with an INAH-led field check.

Encoder Disaggregation in vLLM: Scaling Multimodal Serving with EPD

vLLM details its Encoder/Prefill/Decode (EPD) disaggregation work, including metadata-only requests that let downstream instances reuse remote embeddings without repeating media preprocessing, a dedicated encoder-only Model Runner V2 path, and remote embedding transfer via NIXL and Mooncake. It also covers cross-encoder output reuse through Mooncake Store and orchestration support in vLLM Router, SMG, and llm-d, with core capabilities in vLLM v0.32.0 and DeepSeek V4.1 Flash metadata-only support on main via #62802. A DeepSeek V4.1 Flash benchmark on GB200 found EPD delivered 6.6–9.5% higher per-GPU throughput and 44–47% lower P90 TTFT than PD at concurrency 64–128, but lower per-GPU throughput at concurrency 16.

Experience the art and legacy of Dr. Esther Mahlangu

Google Arts & Culture has launched the first online retrospective of South African artist Dr. Esther Mahlangu, timed to her 91st birthday, featuring more than 45 curated stories and an archive of her art, life and legacy. Curator Nontobeko Ntombela describes using AI, including Veo, to visually recreate Mahlangu's undocumented memories from age 10 until her first documented works in 1980, in collaboration with Google DeepMind, the Google Arts & Culture Lab and Domestic Data Streamers. The exhibition was curated with partners including Art of Contemporary Africa, Iziko Museums of South Africa, the British Museum, the Jean Pigozzi Collection, the National Museum of Women in the Arts and Serpentine, and is available at goo.gle/esther-mahlangu.

MBZUAI’s IFM Releases PAN 2, an Open 29B World Model for Physical AI

MBZUAI's Institute of Foundation Models released PAN 2, an Apache 2.0 world model that takes a scene and text action and simulates the outcome as video, shipping weights, training code and inference code. It pairs an LLM planner with a 29B MMDiT diffusion transformer trained from scratch, outputting 704×1248 video at 24 fps up to 20 seconds. IFM reports 34.0% full-plan completion on 100 multi-step tasks, leading six tested models, while trailing MiniMax-H3 on single actions and human preference.

Seedream 5.5 is now available via Partner Nodes

Seedream 5.5 is now available in ComfyUI via Partner Nodes as a multimodal model that generates from text and edits from images. The release adds multi-turn localized editing, native transparent RGBA output, poster and infographic layouts with readable text, more realistic portraits, and prompt-based lighting control, all behind a single node. Partner Nodes run on credits, and users can load the Seedream 5.5 template from the Templates panel or use Comfy Cloud.

llm-gemini 0.36

Gemini 4 Carbon, announced two weeks ago for "a subset of cyber defenders", is now available to everyone at launch prices of $2/million input and $10/million output, scheduled to double to $4/$20 on December 1st. The post shows three pelican test images for low, medium and high thinking efforts (defaulting to medium) and an Antigravity IDE test producing a Three.js animation of Yonkers as a mech stomping around Manhattan, Queens and Brooklyn using OpenStreetMap data. An update notes a YouTube demo in which Carbon's /swarm feature built an open-world 3D game exploring Google's Mountain View campus as a pelican riding a GBike.

Multimodal Knowledge Agent (MKA) benchmark: testing knowledge work capabilities with complex visual files

Artificial Analysis partnered with Google DeepMind to launch the Multimodal Knowledge Agent (MKA) benchmark, comprising 200 private real-world visual tasks across 10 industries in which agents work from dense visual files like architectural drawings, engineering schematics, and clinical scans to produce deliverables such as slide decks, spreadsheets, PDFs, Word documents, and marked-up images. Tasks are graded against expert rubrics averaging ~70 weighted criteria split into Regression and Climb checks, with the headline Climb WSR score, run in the open-source Stirrup agent harness with up to 250 turns and graded by GPT-6 Luna (medium) as an agentic judge. Claude Fable 5.2 (max) leads with 57% Climb WSR, followed by Gemini 4 Carbon (high) at 54% and Claude Opus 5.5 (max) at 53%, while Claude Fable 5.2 averages $31.97 per task versus $6.29 for Gemini 4 Carbon and $1.11 for GPT-6.1 Sol.

Introducing Scribe v3 and v3 Live, our most accurate speech to text models

ElevenLabs launched Scribe v3 for recorded audio, which makes 27% fewer word errors than Scribe v2 across 102 languages and has the lowest diarization error of any model tested, and Scribe v3 Live for realtime audio, which makes 43% fewer semantic word errors than Scribe v2 Realtime and adds speaker diarization and improved turn detection. Scribe v3 Live is available today in ElevenAgents, and both models are available in ElevenAPI, where switching from Scribe v2 is a one-line change to the model ID (scribe_v3 or scribe_v3_live). Both run in the US by default with isolated EU, India, and Singapore environments, Zero Retention Mode, and SOC 2, ISO 27001, PCI DSS Level 1, and HIPAA-eligible coverage.

Hollywood Has Flipped on GenAI, 'If You Don't Have an AI Strategy, Don't Bother'

Weta FX CEO Daniel Seah said Hollywood studios went from barring any generative AI use in early 2024 to telling vendors since the third quarter of this year that "if you don't have an AI strategy, don't bother to work with us." Weta signed a partnership with Taiwan's TAICCA at the Taiwan Creative Content Fest, and is working with Taiwanese studio Moonshine on artist-layer tools, while CTO Kimball Thurston said Weta uses outside video models such as Kling AI and ByteDance's Seedance where clients permit, hosted outside China for security reasons. Seah said Weta does not intend to replace artists or employees with AI, and the deal will have TAICCA and Taiwan's Ministry of Culture identify young talent for possible Weta internships in Wellington, Melbourne and Vancouver.

Best OCR Model for Scanned Documents, Measured

The author tested 15 vision models plus Mistral OCR and Tesseract on eight synthetic scanned pages scored by character error rate and 264 key-value checks, finding Claude Opus 5.5 most accurate (262 of 264 at $16.42 per 1,000 pages) and Mistral OCR the best value (258 of 264 at $2.02). The post documents failure modes such as fabricated card digits, wifi passwords and server names on a faded receipt, and provides code for transcription, disabling reasoning, cross-checking two models from different labs, and using the mistral-ocr engine.

Photoshop, Lightroom, and Premiere are making it easier to remove things

Adobe announced at its Max event new features including a beta Generative Remove tool in Premiere that erases unwanted objects on the editing timeline, an updated Remove Tool in Photoshop, and desktop blemish removal in Lightroom. Photoshop also gains Quick Compare and Relight, Lightroom adds Open Eyes in public beta, and Firefly gets new video and image tools plus Quick Actions now generally available.

Adobe is going all in on agentic AI

At its Max conference in Miami, Adobe announced that third-party AI agents from providers like OpenAI, Google, and Anthropic will be able to access and modify Creative Cloud projects through local Model Context Protocol integrations, with a private beta planned for early 2027. Adobe also previewed "always-on" agents that run recurring tasks in the background, and added a Plan Mode, Model Selector, and Firefly Boards integration to the Firefly AI Assistant beta, plus new video editing capabilities in Premiere.

Adobe brings AI agents and new creative tools to Photoshop, Premiere, and Firefly

At its MAX event in Miami, Adobe announced new AI features across Photoshop, Premiere, Illustrator, InDesign, Lightroom, and Firefly, including a private-beta Firefly prototyping tool called Builder, an always-on proactive agent, and a local Model Context Protocol (MCP) that exposes Creative Cloud capabilities to chatbots like ChatGPT, Claude, and Gemini. Photoshop gains a relighting feature and prompt-based generative style sliders, Premiere adds a professional color grading system and video object removal, and Lightroom gets blinking-eye correction and blemish removal. Adobe VP Deepa Subramaniam told TechCrunch the company is still deciding which capabilities to expose through the MCP and will rely on user feedback.

We’re bringing Adobe’s creative and productivity tools closer to your everyday Google apps.

At Adobe MAX, Google announced an expansion of its partnership with Adobe that will bring Adobe's creative and productivity tools into Google products, and Google's frontier AI models into Adobe's apps. Planned integrations include a direct pathway from Google Drive and Gmail into Adobe Acrobat with an "Edit with" entry point in file previews, expanded Adobe connectors in Gemini and Gemini Enterprise, Adobe creative tools in AI Mode in Google Search, and Adobe's CX Enterprise Coworker agent in Gemini Enterprise. Adobe and Google DeepMind will also explore new editing capabilities for creative professionals, starting with an image model for Photoshop that could enable near-real-time edits.

Show HN: AI Vision Index, private tests of how well 36 AI models see

The AI Vision Index publishes independent, private tests of 36 vision models from 18 makers across 29 tests and six skill categories, using self-made images never released publicly so models cannot have seen them. GPT-6 Astra medium leads with 83.7 ±2.0 out of 100 at $38.32 per 1,000 answers, followed by GPT-6.1 Sol medium (81.3) and Claude Fable 5.2 (80.7), while local models such as LFM2.5-VL 1.6B score 12.4. Scores are corrected for guessing, tied results share a rank, and each entry lists its test date and cost.