Model launchesThinking Machines Lab unveils TML-Interaction-Small for real-time, parallel multimodal interactionThinking Machines Lab introduced TML-Interaction-Small, a multimodal conversational system designed to process audio, video and text in parallel and generate outputs concurrently using…4 min read
Model launchesGoogle revamps Search, commerce and work with always-on AI agents and unified commerce protocolsAt I/O 2026 Google unveiled wide-ranging changes to Search, Workspace and shopping: the search box will now process multimodal inputs and deliver AI summaries, while persistent AI agents will monitor topics and act on users’ behalf.5 min read
Model launchesThey will build a new, significantly larger model using ten times more computation in partnership with SpaceXAISpaceXAI and the authors of the statement announced that they are training a completely new, significantly larger language model from scratch, using ten times more total computation.1 min read
Model launchesComposer 2.5 Released: Improved Language Model for Long TasksThe developer introduced the Composer 2.5 model, which is smarter, handles long-running tasks better, and follows complex instructions more reliably.1 min read
Model launchesOpenAI releases GPT‑Realtime‑2 speech‑to‑speech model with adjustable speed–reasoning tradeoffOpenAI launched three new audio models in its Realtime API: GPT‑Realtime‑2 (speech‑to‑speech with configurable reasoning effort), GPT‑Realtime‑Translate (speech translation), and GPT‑Realtime‑Whisper (speech transcription).4 min read
Model launchesGemini Intelligence: expected to debut first on Samsung foldables and Google Pixel devicesGoogle unveiled the Gemini Intelligence AI assistant at the Android Show: I/O Edition event on Tuesday.2 min read
Model launchesOpenAI and Thinking Machines Redefine Human–Machine InterfacesOpenAI has launched GPT‑Realtime‑2, a GPT‑5‑class voice model designed for live audio reasoning, translation, transcription and tool use with a 128K context window.2 min read
Model launchesVoxtral TTS: lightweight 4B model rivals ElevenLabs, adds emotion steeringVoxtral TTS, a 4-billion-parameter text-to-speech model, outperformed ElevenLabs Flash v2.5 in human preference tests for naturalness and matched ElevenLabs v3 quality while offering emotion-aware control.3 min read
Model launchesThinking Machines Lab unveils a real-time Interaction Model led by Mira MuratiThinking Machines Lab — led by former OpenAI CTO Mira Murati — announced its first Interaction Model after 18 months of development.2 min read
Model launchesLeaked Gemini Omni shows improved on-screen text stability in video but mixed visual qualityA leaked build of Google’s Gemini Omni, surfaced via the Gemini app, reveals a native video model capable of generation and chat-driven editing.2 min read
Model launchesByteDance integrates Seedance 2.0 into CapCut, bringing multimodal video generation to millionsByteDance has integrated its multimodal video generator Seedance 2.0 into the CapCut video-editing app and widened availability to paying users across many regions.4 min read
Model launchesNew real-time translation model available via APIA technology developer announced the availability of a new real-time translation model; according to the statement, the model can be tried via the API from today.1 min read
