Translate
Live intelligence

Latest AI updates

533 updates
AllModelsProductsIndustryResearchTutorialOpinion
Industry

Anthropic Reports Q3 Revenue of $16.8B Ahead of IPO

Anthropic told investors its quarterly revenue rose to about $16.8bn in Q3, up from $11.5bn in Q2 and roughly $4.8bn in Q1, according to an FT report cited by Simon Willison. The company also disclosed that adjusted operating margins excluding stock-based compensation were 11%, and that it earned $1.70 for every $1 spent on computing power. Anthropic said it materially reduced "distillation" usage by rival labs in Q3, a practice that had accounted for 5% to 7% of revenue.

Simon Willison2d agoScore 75
AnthropicAI industry financialsmodel distillation
Models

Mistral AI Introduces Timestral Forecasting Models

Mistral AI has announced Timestral, a family of probabilistic forecasting models released in four sizes ranging from 5M to 209M parameters. The company describes the models as state-of-the-art, CPU-efficient and zero-shot ready. The announcement provides no benchmark details, availability information or pricing.

Mistral AI2d agoScore 76
Mistral AItime series forecastingefficient AI models
Products

Amazon's Bee launches two new personal AI devices, starting at $75

Amazon-owned Bee has released its first AI devices since the acquisition: an updated wearable and a credit-card-sized recorder that attaches to smartphones. Both devices can summarize conversations, create to-do lists, and integrate with Alexa and Amazon shopping, with prices starting at $74.99.

TechCrunch AI2d agoScore 70
AmazonBeeAI hardware
Products

Security review surfaces privacy risks in Eiliko AI companion robot

The Verge reports that the Eiliko AI companion robot, made by Energize Lab, sends conversations to Chinese servers where stored data could be accessed or the bot's personality hijacked. In a ten-minute test conversation, the robot told the reporter their secrets were safe. The article is a truncated security and privacy review of a consumer AI toy rather than a product or model launch.

The Verge AI2d agoScore 70
EilikoEnergize LabAI companion robots
Industry

Anthropic's Q3 revenue reaches $16.8B as growth slows ahead of planned November IPO

Anthropic reported $16.8 billion in third-quarter revenue, a 45 percent gain that falls well short of the 140 percent growth seen in Q2. With a planned November IPO approaching, some investors reportedly expect the company's valuation to drop below $1.5 trillion. Slowing growth and a tough market are tempering expectations for what the article calls the AI boom's biggest Wall Street test yet.

The Decoder2d agoScore 71
AnthropicAI industry fundingIPO
Models Editor selected

Meet Grok Voice 3.0: SpaceXAI Voice Agent Model Ranks #1 on Artificial Analysis

SpaceXAI released Grok Voice 3.0, an API-only real-time speech-to-speech voice agent model. It ranks first on the Artificial Analysis Speech-to-Speech Index with a score of 86.9 and achieves a 0.81-second time to first audio. The model maintains the prior $0.08-per-minute pricing and offers OpenAI Realtime API compatibility for developers.

Why it matters: Offers developers concrete benchmark data and deployment details for a top-ranked real-time voice agent model.

MarkTechPost2d agoScore 81
Grok Voice 3.0Artificial Analysis Speech-to-Speech Indexvoice agents
Products

Datalab Releases OmniParseBench: An Open OCR Benchmark With 16,288 Pass/Fail Tests

Datalab has released OmniParseBench, an open OCR benchmark that scores document parsers on 16,288 yes/no tests across 2,937 pages from 2,343 documents in 92 languages. The page data, tests, code and offline scorer are public, with results sliceable by tags on five content axes and per-test provenance. At launch, Datalab reports Gemini 3.8 Flash leading at 92.9, Datalab accurate at 92.7, and Claude Sonnet 5.5 posting the top text and tables scores. The benchmark installs via uv pip, requires Pyth

MarkTechPost3d agoScore 75
OmniParseBenchOCR evaluationDatalab
Models Editor selected

Fish Audio Releases Drama 3, a 35B Open-Weight MoE Text-to-Speech Model

Fish Audio released Drama 3, a 35B mixture-of-experts text-to-speech model with about 3B active parameters per token, built on a Qwen3.5-35B-A3B backbone. It accepts plain-language acting directions, paired tags, and up to 16 speakers per request across 83 languages, with weights on Hugging Face and availability via the Fish Audio API. Self-hosting requires a single NVIDIA GPU with at least 48 GB in FP8 using SGLang-Omni, which supports only Hopper GPUs. It scores 93.2% overall on Artificial Ana

Why it matters: Developers evaluating open-weight TTS get concrete architecture, hardware, benchmark, pricing, and licensing details to decide between self-hosting and the hosted API.

MarkTechPost3d agoScore 83
Fish Audio Drama 3open-weight TTS modelsQwen3.5-35B-A3B
Industry

Eight US states sue OpenAI over ChatGPT child safety claims

Eight US states—Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana—have filed separate lawsuits against OpenAI, alleging it harmed children and misled customers about ChatGPT's safety. The complaints claim OpenAI prioritized profits over safety and violated local consumer protection laws by not warning users about dangers. The suits reference prior litigation from families alleging the chatbot played a role in harm to teens and adults.

The Verge AI3d agoScore 78
OpenAIChatGPTAI regulation
Industry

40 State AGs Intervene in Character.AI Bankruptcy Over User Data Protections

New York Attorney General Letitia James has revealed a multi-state investigation into Character.AI, with 40 bipartisan attorneys general filing a notice of interest in the company's Chapter 11 bankruptcy case. The AGs want consumer data protections addressed in any sale or transfer of assets. Character.AI announced it was filing for Chapter 11 and selling some assets to Disney.

The Verge AI3d agoScore 75
AI regulationdata privacyCharacter.AI
Research

AI-Designed Proteases Cut Specific Proteins, Including Neurological Disease Targets

Researchers used AI-based protein design tools to take the catalytic core of two types of protease and replace the rest of the protein, producing engineered enzymes that cut specific proteins in half. They demonstrated the enzymes by cutting proteins associated with neurological diseases. The work short-circuits the natural diversification of protein-digesting enzymes, allowing the resulting proteases to be tuned to target chosen proteins.

Ars Technica AI3d agoScore 75
AI protein designproteasesneurological disease
Tutorial

AWS shows self-improving agentic document processing on Bedrock AgentCore

AWS published a tutorial describing a multi-agent intelligent document processing system built on Amazon Bedrock AgentCore. Eight specialized agents classify, extract, validate, and save reusable instructions for new document formats using AWS services including Amazon Textract, S3 Vectors, Aurora DSQL, Lambda, and AgentCore Gateway. The post provides a deployable sample repository, sample purchase order flows, estimated costs, and guidance for adapting schemas, validators, and prompts to other

AWS Machine Learning Blog3d agoScore 72
Amazon Bedrock AgentCoreAWSDocument AI
Research

Microsoft open-sources machine learning tool for subseasonal weather forecasting

Microsoft researchers say they have developed a machine learning method that improves weather forecasts two to six weeks ahead, a horizon where physics-based and existing AI models typically lose reliability. The findings were published in Nature, and the tool has been released as open-source software. Microsoft joins Google, Nvidia and others working on AI weather forecasting, with potential users including farmers and emergency responders.

The Verge AI3d agoScore 70
MicrosoftAI weather forecastingopen-source research
Models Editor selected

Liquid AI Releases LFM2.5-24B-A2B: A 2B-Active MoE for On-Device Agents

Liquid AI released LFM2.5-24B-A2B, an open-weight 24B-parameter Mixture-of-Experts reasoning model with about 2B active parameters and 131,072-token context, targeted at local agent loops. It runs via GGUF/MLX on laptops and mini PCs, with cited speeds of 184 tokens/s on Apple M5 Max and 114 tokens/s on AMD Ryzen AI Max+ 395, plus single NVIDIA B300 serving. Liquid AI reports it leads or matches larger MoEs on several agentic, tool-use and instruction-following benchmarks, while trailing Qwen3.6

Why it matters: Provides a concrete open-weight release with on-device performance, agentic benchmark comparisons, licensing and deployment details useful for evaluating local agent models.

MarkTechPost3d agoScore 82
Liquid AILFM2.5Mixture-of-Experts
Models Editor selected

Google Releases Gemini 4 Argon API With 1M Token Output Limit

Google DeepMind made Gemini 4 Argon generally available to developers, enterprises, and paid Gemini subscribers through the Gemini API, AI Studio, and Gemini platform. Argon is a proprietary sparse mixture-of-experts frontier model with native text, image, audio, and video input, and it supports up to 1M tokens for combined input and output, including responses up to 1M tokens. Introductory API pricing is $2 per million input tokens and $10 per million output tokens, rising to $4/$20 after Novem

Why it matters: This is a major frontier-model API release with a notable long-output capability, concrete pricing, benchmark results, and caveats that matter to developers choosing between leading models.

MarkTechPost3d agoScore 89
Gemini 4 ArgonGoogle DeepMindFrontier AI models
Industry

Anthropic Built Its Own Source Control System to Handle AI-Driven Code Velocity

Anthropic has built a custom replacement for Git to cope with a sharp rise in code changes that it attributes to AI coding agents. According to the piece, Anthropic averaged around 2,500 merged changes per day in May 2026 and over 30,000 per day by September. The new source control system is called Scion, joined by a code review system called Antique, and was built by a team of seven with the migration from Git completed in five months. Scion intentionally omits offline support because Claude is

Simon Willison4d agoScore 75
AnthropicScionversion control
Models

Meta Launches Muse Wonder and Muse Wonder Full: Benchmarks, Pricing and How to Use It

Meta has launched Muse Wonder and Muse Wonder Full, frontier models aimed at long-running agent work such as coding, legal, finance, and computer use. Muse Wonder is available via Meta Model API and Muse Code at $4/$20 per 1M tokens, while Muse Wonder Full is limited to paid Muse Code plans starting at $50/month. Both offer 1M-token context and text, image, video, and audio input, and run only on Meta’s cloud. In Meta’s published benchmark table, Muse Wonder Full leads 6 of 14 rows and Muse Wond

MarkTechPost4d agoScore 77
MetaMuse Wonderfrontier models
Models Editor selected

llm-gemini 0.35 adds support for Gemini 4 Argon

Simon Willison released llm-gemini 0.35, adding support for Google's new Gemini 4 Argon model, available via the Gemini API and a $100/month subscription. At launch, Argon is priced at $2 per million input tokens and $10 per million output tokens, with a planned price doubling on December 1st. Artificial Analysis data shows Argon costing $2,407 to run their benchmark, placing it close to GLM-5.3 and cheaper than recent models from OpenAI, Anthropic, and Meta. The post also includes SVG generatio

Why it matters: Provides developers with timely, concrete details on a new Gemini model's pricing, benchmark cost, and real-world performance for model selection.

Simon Willison4d agoScore 83
Gemini 4 ArgonGoogleLLM pricing