To get access to the latest Gemini models, including the most recent experimental release that's currently at the top of the @lmsysorg leaderboard, use @googleaistudio. You can also turn on features like code execution; use extended context (2M+); and turn off safety settings:

model scores 94% on your eval you open the traces it's been reading the answers out of the filenames

very full day! ✨ started in san jose speaking at @WeAreDevs world congress (yes, the giant duck was there for moral support) then up to @google sf for the ai for social good hackathon with @valeriemchapman: multiple rooms full of women turning ideas into real prototypes with @googleaistudio i got to talk with teams as they worked through their concepts, and the care and attention they put into their pitches was very inspiring to see - everything from safety features in @googlemaps to nudges to build community in @partiful thank you to everyone who came to the talk (virtually and in person!), and to valerie and team for building a space like this 💙

...and more people keep coming! 🤯 Great conversations, everything from long horizon audio tasks (things like customer service phone calls), to on-device audio running on robots, to asynchronous meetings managed by agents. And the absolute coolest Lauryn Hill shirt! ♥️🙌

Packed house for the @GoogleDeepMind @GoogleAIStudio Audio Event tonight! 🎤 ✨ Come learn about our new text-to-speech models, which include voice personalization (!!), as well as the updates we've made for live translation, live transcription, and Gemini Live.

🎛️🎶 Did you know that you can use Gemini 3.8 Flash in @GoogleAIStudio to generate MIDI tracks, based on a text description, an input video, or an image -- and then directly import them into Garage Band (or another music creation platform)? The example below is just from the prompt: "Create a midi file for a knight character about to fight the final boss in a game scene."

+100, the half-life of "that's impossible!" -- for everything from mathematics to medical research to robotics and space travel -- is down to about four days right now. There’s nowhere else I’d rather be.

This is why open source matters so much! When community-driven tools like @cline can give every developer free access to ~300 tok/s and 1M context, the barrier to building truly drops to zero. No gatekeeping, no massive API bills - just pure enablement for builders everywhere (especially students). Go create something cool! 🛠️✨

“MVP carries four Google tensor processing units... Enough to answer simple Gemini queries from orbit.” Launching actual @GoogleCloud TPUs into low Earth orbit on Oct 1! 🛰️✨ Designing thermal management and radiation tolerance without conventional convection or fans is such incredible systems engineering.

Pretty cool night! Met @DynamicWebPaige and learned from bunch of cool researchers at @PatronusAI on how they think about evals. thanks for the invite @gracegongGG

🙌 Make it loud, see y'all tomorrow!

🙌 All things data at @PatronusAI and @gracegongGG's panel session on evals tonight! Everything from SpeedrunBench (how fast an agent can finish a game) to benchmarks for world models and even evals for agent actions in spreadsheets with embedded formulas and charts (!!). 📊📈

spent the last decade+ helping folks write software that does things on a screen; the next decade is going to be all about software that picks things up discussion panel on robotics + physical ai at dirtyjobs.ai later today. every company in the future is a robotics company!

🙌 Open models have come to the @Antigravity SDK! Test out @GoogleGemma 4 today:

...and it even works in other languages! 🇩🇪 Test out the new text-to-speech models in @GoogleAIStudio today: