@melvinjohnsonp
Post-Training lead, Gemini, @GoogleDeepMind
AI
The v1.0 benchmark spans 11 domains and 206 tasks, with a score that averages performance across domains.
AI
Gemini 4 Carbon's early evaluations claim a new state of the art on the Wide-And-Deep-Research benchmark.
AI
A post says Carbon scored 88.2% on SEC-bench Pro and is rolling out first to cybersecurity specialists.
AI
One poster credits the switch alone for an improvement and says more work remains on the harness.
AI
Google DeepMind says the Antigravity creation animates orbiting electrons, atomic data, and light spectra.
XAI
shoyer praised Argon's writing but said it doubled down on what they called a lie.
AI
Perplexity says Pro and Max subscribers have access; a separate post says the model is available to all users.
AI
A post calls GDM’s results impressive and notes a “backwards bend” in action efficiency.
AI
The post describes a discussion of “how we got here” and the path ahead for Gemini 4 Argon.
AI
A user says Gemini Argon built a simulation and trained the Apollo humanoid’s throws on the real robot.
XAI
A post sharing a conversation with Amplitude, Deloitte and Rakuten leaders says the model scored 90%+ on data visualization evaluations.
AI
Google says Argon is available through its paid API tier and to Pro and Ultra subscribers in Gemini and Search.
YouTubeAI
Google announces a 1 million-token output limit, up from 64,000, for its new model. Initial access goes to trusted cyber defenders through the Fairwind Program.
AI
New Flash models target agentic workflows and cybersecurity tasks with API and enterprise access.
AI
Google engineers note rapid adoption and positive feedback for the release.
AI
Google's model ranks first on ArtificialAnlys benchmark for quantitative analysis tasks.
AI
Google's new model improves coding accuracy and agentic tasks at reduced cost.
