Google releases Gemini 4 Hydrogen, built for fast coding and agent tasks
Google's new model has a context window of 1 million tokens, at $1.50 for each million tokens of input and $7.50 for each million of output.
Models
New AI models from the labs, and what each one is built to do.
MiniMax releases open-weight H3-Image model for 4K image generation
MiniMax says H3-Image can produce a 4096 by 4096 pixel image in less than 10 seconds and is available as open weights and through its API.
Meta introduces Muse Ember, a small, low-cost model for repeated agent tasks
Meta prices Muse Ember at $0.10 per million input tokens and says it matches Claude Haiku 5.5's best index score at roughly one-third the cost.
NVIDIA contributes OpenShell agent-security runtime to CNCF, adds three Kubernetes tools to DSX
NVIDIA says it is giving its OpenShell agent-security runtime to the CNCF and releasing three open source Kubernetes components for its DSX AI Factory Platform.
OpenAI offers ChatGPT Enterprise for purchase through AWS Marketplace
OpenAI says organizations can now buy ChatGPT Enterprise through AWS Marketplace, with purchasing and billing handled through their existing AWS relationship.
Google DeepMind introduces Gemini Robotics 2.5, reporting 96% success in Mercedes-Benz humanoid pilot
Google DeepMind says its new robot-control model is rolling out to partners and reports a 96% success rate in an Apptronik humanoid trial on Mercedes-Benz production lines.
Google opens Gemini 4 Carbon to paying developers, enterprises and AI Ultra subscribers
Gemini 4 Carbon, announced 28 October and first released to cyber defenders, is now open to paying API users and rolling out to AI Ultra subscribers.
NVIDIA introduces GR00T 2, an open robot model built on Cosmos, at CoRL
NVIDIA says GR00T 2 learns dexterous tasks from a few demonstrations; it also introduced Isaac Lab 3.0 and two open benchmarks at CoRL.
Anthropic releases Claude Fable 5.2 and Mythos 5.2 at a lower price
Anthropic says its 5.2 models generate output as much as 30% quicker than the 5.1 models, with input at $8 and output at $40 per million tokens.