Introducing Gemini 4 Hydrogen
Gemini 4 Hydrogen is a faster, cheaper Gemini 4 tier rolling out across AI Studio, the Gemini API, the Gemini app, Search and Antigravity, priced at $1.50 per million input tokens and $7.50 per million output tokens.
First-party releases and research from Google DeepMind. Headlines and excerpts link to the original articles.
Gemini 4 Hydrogen is a faster, cheaper Gemini 4 tier rolling out across AI Studio, the Gemini API, the Gemini app, Search and Antigravity, priced at $1.50 per million input tokens and $7.50 per million output tokens.
Google DeepMind has announced AlphaGenome 2, a conditional DNA sequence model that predicts variant effects for individual cell types and states rather than bulk tissues. It identified causal variants on the TenK10K immune dataset at 30% versus 16% for the original model.
Google DeepMind released Gemini 4 Carbon, a frontier model positioned above Gemini 4 Argon for deep multi-step coding, scientific research and cybersecurity work, priced at the same introductory rate of $2 per million input tokens and $10 per million output tokens.
Google DeepMind released SatWeave and PARDIG, two datasets used in WeatherNext 3, as open historical archives and live feeds on Google Cloud Storage, with PARDIG historical catalogs also in Earth Engine.
Google DeepMind released Gemini Robotics 2.5, a vision-language-action model that improves success rates by 30% to 100% over prior baselines across 200 tasks and helped an Apptronik Apollo 2 humanoid achieve a 96% success rate on a Mercedes-Benz production-line kitting pilot.
Google DeepMind is working with the UK government on AI intended to shorten planning assessment times for housing, part of a broader partnership between the two announced in a post titled "Unlocking UK house-building with AI-accelerated planning."
Google DeepMind announced Gemini 4 Carbon, a frontier model it says leads on cybersecurity benchmarks and tops the v5.0 Artificial Analysis Intelligence Index at 59, with initial access limited to a subset of cyber defenders.
Google DeepMind announced Genie 4, a natively multimodal real-time world model that generates photorealistic interactive simulations with sound and dialogue, now in a research preview for trusted testers. Consumer and developer access is planned to start with Google AI Ultra subscribers in Project Genie and early access partners of the Gemini API and Google AI Studio.
Google DeepMind demonstrated that its agentic platform Google Antigravity, paired with the ProtNLM function-prediction model, surfaced a novel antimicrobial peptide candidate from Streptococcus pseudopneumoniae; the lead is now in wet-lab validation. Google also introduced a Biomolecular Structure Viewer embedded in Antigravity.
Google DeepMind is broadening access to Gemini 4 Argon, its frontier model with strong cyberdefense abilities, after a limited Fairwind Program phase in which Wiz used it to find a critical hospital software bug. The company detailed new safeguards for misuse, misalignment and prompt injection.
Google DeepMind released Gemini 4 Argon, a frontier model available globally via the Gemini API, AI Studio and Google subscriptions, priced at an introductory $2 per million input tokens and $10 per million output tokens. Google says it ranks first on the Vals Index and outperforms other frontier models on domain benchmarks.
Google DeepMind released EmbeddingGemma 2, an open lightweight multimodal embedding model with 740 million parameters under an Apache 2.0 license. It maps text, images, audio, and video into a unified embedding space and is built on the Gemma 4 architecture.
Google DeepMind announced Gemini 4 Argon, a frontier model rolling out to trusted cyber defenders through its Fairwind Program before wider release. It offers a 1 million token output limit, introductory pricing of $2 per million input tokens and $10 per million output tokens, and reported benchmark results across software engineering, knowledge work, and cybersecurity.