Google DeepMind
Google DeepMind is Alphabet's AI division, home of the Gemini model family, the Nobel-recognized AlphaFold and the Veo video generator. Through Google Search, Android and the cloud, its models reach billions of users. This page gathers all our coverage of Google and DeepMind: models, research and products.

Google launches Gemini Decision 1: video-based decisions at $0.10 per million tokens
How do you route requests containing photos and videos without lengthy LLM responses? Gemini Decision 1 selects from predefined options and returns probabilities.

MWS cut AI agent startup from 37.6 to 16.8 seconds: a harness teardown
Users waited almost 40 seconds for a corporate agent's first reply. Here is where the time went, how MWS cut it by more than half, and the pitfalls along the way.

DeepMind Releases Gemini Robotics 2.5: Mercedes Robots Assemble Parts with a 96% Success Rate
Gemini-powered robots are kitting parts at Mercedes plants in Europe and succeed 96% of the time. Alongside the model, DeepMind has opened a benchmark that tests…

Claroc: 95% of 37 Thousand Apps Built in Grok Are Open to Anyone Without a Login

Google releases Gemini 4 Carbon: tops three agentic benchmarks at less than half the cost

Google Opens Gemini 4 Carbon to Developers: $2 per Million Tokens and a Swarm of Agents

Anthropic Releases Claude Fable 5.2: 1,000 Agents Formalized Fermat's Theorem in 30 Hours

DeepMind’s AlphaProof Code verified Signal’s cryptographic protocol in 4 weeks
Formal verification was considered the preserve of a handful of experts and took years. DeepMind’s agents wrote 6 million lines of Lean proofs for Signal’s code—and…

Anthropic Limits Claude and the API to 185 Countries: What Changes for Teams
Claude, Claude Code and the API now operate on an allowlist of 185 regions. Business trips, employees from excluded countries and even indirectly controlled…

Anthropic Releases Model Selection Guide: Haiku 5.5 Saves Tokens on Routine Work

Google launches Gemini Decisions API: up to 1000 checks in a single request

Gemini Robotics 2.5 helped Apollo 2 sort Mercedes parts with a 96% success rate

Anthropic releases Claude Fable 5.2: 65 points in the ranking, but $26.50 per task

Google releases Gemini 4 Carbon: a 1 million-token window for complex debugging
When AI fixes one file and breaks another, the entire chain needs checking. Google has made Gemini 4 Carbon available for these tasks, with paid access, extended…

Meta Muse Ember Matches Claude Haiku 5.5 on Score at One-Third the Cost per Task
Meta's new compact model scores 36 on the Artificial Analysis index, the same as Claude Haiku 5.5, but a single task costs 14 cents instead of 41. There is a catch…









