Meta Unveils New Features for AI Agent Muse
Meta announced new capabilities for its AI agent Muse, set to integrate with smart glasses and expand into daily tasks by October 2026.
Model Release
576 articles
New AI models as they ship — large language models, multimodal systems, and open-weight releases. Capabilities, context windows, pricing, and availability, summarized so you can compare what is actually new.
LATEST NEWS
-
-
Black-forest-labs Launches FLUX 3 Action, Open-Weights World Action Model
Black-forest-labs released FLUX 3 Action, a 7B parameter open-weights world action model, achieving 42.92% success on RoboLab benchmark, outperforming 16B Cosmos 3 Nano policy with 36.8%.
-
Airbnb Expands Access to GPT-6 Astra and OpenAI Models
Airbnb announced broader access to GPT-6 Astra and OpenAI frontier models on September 23, 2026, as part of a new agreement with OpenAI.
-
Meta's AI Agent Muse Gains 500,000 Users in First Week
Meta's AI agent Muse attracted over 500,000 users in its first week, including 250,000 daily active users, after its September 8 launch.
-
Anthropic Engineer Explains Why Claude's Writing Got Worse Despite Model Improvements
Anthropic engineer Jackson Kernion said Claude's writing quality declined after Opus 4.6, despite the model becoming smarter.
-
Google Unveils Flash TTS Models for Custom AI Voice Creation
Google released two new text-to-speech models, Flash TTS and Flash-Lite TTS, supporting over 100 languages and enabling voice creation from text descriptions.
-
ChatGPT Voice Uses GPT-6 Models for Email and Slack Access
OpenAI launched ChatGPT Voice with email, calendar, and Slack access on September 23, 2026, marking its first major update since the 2024 release.
-
NVIDIA Unveils Nemotron 3 Diarization for Speaker Identification
NVIDIA's Nemotron 3 Diarization model ranks #1 in VoiceArena's Diarization-Bench leaderboard with a 14.72% Diarization Error Rate (DER).
-
Basis Releases Conversations 1500 Dataset for Multilingual Research
Basis AI released Conversations 1500, a dataset containing 1,502 hours of multilingual, multi-party conversation, on September 23, 2026.
-
DeepMind Launches Gemini 3.8 Text-to-Speech Models
DeepMind released Gemini 3.8 text-to-speech models on September 23, 2026, offering advanced audio generation capabilities for creators and developers.
-
OpenAI Introduces GPT-6 Sol and Luna with Half the Price
OpenAI launched GPT-6 Sol and Luna, cutting token prices by 50% compared to GPT-5.6 models, while maintaining similar performance levels.
-
Meta Admits Muse's Similarity to OpenClaw Isn't a Coincidence
Meta acknowledges Muse's design was inspired by OpenClaw, with shared files and features, as the AI tool climbs to No. 1 on the U.S. App Store.