Model launchesGoogle introduces Gemini 3.5 Transcribe, an AI model for structured transcriptions and on‑page dictationGoogle unveiled Gemini 3.5 Transcribe, an audio-focused AI model designed to convert unstructured spoken language into formatted text and to support voice-driven editing.2 min read
Model launchesQwen3.8-Flash-Next: open multimodal Mixture-of-Experts model previewing Qwen4 architectureQwen3.8-Flash-Next is an open-weights, multimodal Mixture-of-Experts (MoE) model that serves as an early preview of the architecture planned for Qwen4.2 min read
Model launchesAnthropic expands Claude's context window to 100,000 tokensOn May 11, 2023 Anthropic announced that its Claude model's context window has been increased from 9,000 to 100,000 tokens (around 75,000 words).2 min read
Model launchesAlibaba previews Qwen3.8-Flash-Next: a long-context multimodal MoE model with 262k native context and 1M-token scalingAlibaba released weights for Qwen3.8-Flash-Next as a developer preview of the upcoming Qwen4 family.3 min read
Model launchesFormer Meta Researchers Launch Vision AI for Industrial RobotsPerceptron, a startup founded in November 2024 by former Meta FAIR researchers Armen Aghajanyan and Akshat Shrivastava, released Isaac 0.5, a vision model intended to help robots perceive, reason and act in industrial environments.3 min read
Model launchesOpenAI to deploy the Jalapeño model into its computing infrastructure by year-endOpenAI announced that it will begin deploying the Jalapeño model into its own computing infrastructure by the end of the year; this is the first step in a multi-generation roadmap.1 min read
Model launchesInherent unveils Faraday, a 27B-agent that reproduces published experiments by supervising a coding modelInherent, a London AI lab founded by DeepMind alumni, has introduced Faraday, an AI agent built on Qwen 3.6 (27 billion parameters) trained with reinforcement learning to plan and judge scientific experiments.2 min read
Model launchesGPT‑5.6 models integrated into Kiro development environmentThe GPT‑5.6 family (Sol, Terra, Luna) is now available within Kiro, a software development agent that converts high‑level intent into requirements, designs, and executable tasks.2 min read
Model launchesNVIDIA releases 4B Cosmos 3 Edge for on-device robot control on Jetson ThorNVIDIA introduced Cosmos 3 Edge, a 4-billion-parameter omni-model designed for on-device robotic manipulation and real-time control on NVIDIA Jetson AGX Thor.5 min read
Model launchesWaymo expands cheaper Ojai robotaxi to riders in three US citiesWaymo has made its next-generation Ojai robotaxi available to all riders in Los Angeles, Phoenix, and San Francisco, marking a step toward a lower-cost, scalable fleet.3 min read
Model launchesZhipu’s GLM-5.3 narrows gap with Western models after benchmark showing strengthsChinese startup Zhipu (Z.ai) says its latest GLM-5.3 model outperformed Anthropic’s Mythos 5 on a key cybersecurity benchmark and improved in coding tasks, according to the company.2 min read
Model launchesQwen 3.8 27B: a powerful local multimodal model that over-thinks at default settingsAlibaba's Qwen research lab released Qwen 3.8 27B, an Apache‑2 licensed 27B-parameter multimodal LLM that runs locally as a ~17GB file and shows improved benchmarks over prior Qwen versions.5 min read
