W. Ronny Huang

Realtime Gemini models @GoogleDeepMind

[W. Ronny Huang @wronnyhuang]

Controlling antigravity with real-time audio-to-audio is a multi agent orchestration problem. Talk to it and watch it work in the background while staying engaged with you in the foreground. Was lots of fun experimenting and shipping this with @vibe__coder

[Google Antigravity @antigravity]

Voice mode in Antigravity is here. Enabled by the Gemini Live model, you can use your voice to plan, steer code changes, and approve permissions out loud. Have a live conversation across multiple languages, interrupt whenever you need to, set your preferred voice and playback speed, and when you're done, just say "bye" or tap end. Happy building!

[W. Ronny Huang @wronnyhuang]

@kevinhou22 Game changer!

[W. Ronny Huang @wronnyhuang]

genie4 is a highly general visual renderer that can be text-controlled at fine time granularity. orchestrated in the right way with LLMs, this can unlock a lot of useful applications

[W. Ronny Huang @wronnyhuang]

realtime video generation is part of the "interaction model" paradigm, making AGI easier to interface with. earlier this year we launched gemini 3.8 with live avatar, a realtime videogen model based on genie3. happy to have played a part in this next evolution with genie4 🧵

[Sundar Pichai @sundarpichai]

Excited to share Genie 4 today our most realistic and interactive world model:

- steer dynamic environments in real time, shaping events, camera angles, and actions

- generate native audio matched to the world, from realistic background sounds to characters who can converse in real time!

- simulate photorealistic scenes and real-world physics

- create more coherent multi-step events through reasoning from Gemini

We've started to roll out as a research preview to trusted testers. Coming soon to Google AI Ultra subscribers in Project Genie and early access partners via the Gemini API and AI Studio.

[W. Ronny Huang @wronnyhuang]

@petergyang Funny timing, we should have used your generations for our Gemini 3.8 Live Avatar promo video. Our Japanese tutor demo is the last video here: https://t.co/4AlmTMOJIB

[Artificial Analysis @ArtificialAnlys]

Google’s new Gemini 4 Argon equals GPT-6 Astra on the Artificial Analysis Intelligence Index at 60% of the Cost per Task with discounted prices. Google is now back to being one of the top three labs in intelligence achieved

Gemini 4 Argon is @GoogleDeepMind’s first proprietary model above the Flash class in over 7 months. With high reasoning (the highest available), it scores 53 on the Artificial Analysis Intelligence Index, matching GPT-6 Astra (max, 53) and 1 point ahead of GPT-6.1 Sol (max, 52), with gains driven by lower hallucinations and stronger agentic capabilities.

At its current 50% pricing discount and with cache discounts increased to 95%, Gemini 4 Argon costs $1.99 per Intelligence Index task, 60% of GPT-6 Astra (max), but 2.7x GPT-6.1 Sol (max). After the discount ends, this will rise to $3.98 (~1.2x GPT-6 Astra (max)).

Gemini 4 Argon is currently being rolled out to selected users and is not publicly available. The 50% discount is an initial promotion. Google has not yet confirmed the promotion end date

Key benchmarking results for Gemini 4 Argon with high reasoning:

➤ Google returns as one of the top three labs on intelligence: Gemini 4 Argon (high) scores 53 on the Artificial Analysis Intelligence Index, matching GPT-6 Astra (max, 53) and 1 point ahead of GPT-6.1 Sol (max, 52). This is 23 points above Google’s previous non-Flash model, Gemini 3.1 Pro Preview (30) and 12 points ahead of Gemini 3.8 Flash (high)

➤ Launch discounts of 50% make Gemini 4 Argon competitive on Cost per Task: At current discounted pricing, Gemini 4 Argon (high) costs $1.99 per Intelligence Index task, 60% of GPT-6 Astra (max, $3.26) for a comparable level of intelligence. This cost efficiency is driven by lower token prices, rather than reduced token use, with Gemini 4 Argon averaging 62k output tokens per task, compared with 27k for GPT-6 Astra (max). Google has not yet confirmed the promotion end date, but on standard pricing, Cost per Task will increase to $3.98

➤ Stronger agentic performance: Historically a weaker area for Gemini models, Gemini 4 Argon shows improvements across agentic evaluations. It ranks #1 on AutomationBench-AA at 77.5%, 6 points ahead of Claude Sonnet 5.5 (max, 71.3%). On Terminal Bench 4, Gemini 4 Argon achieves 57%, a +53 point improvement from Gemini 3.1 Pro Preview, only behind Claude Sonnet 5.5 (max, 64%), Claude Opus 5.5 (max, 60%) and GPT-6 Astra (59%). On AA-Briefcase, it reaches 1494 Elo. This is driven by a 65% rubric pass rate, the highest we have recorded, but lower Analytical Quality (1576 Elo) and Presentation Quality (1308 Elo)

➤ Lowest hallucination rate among leading models: On AA-Omniscience, Gemini 4 Argon has a 15% hallucination rate, the lowest of any model scoring 45+ on the Intelligence Index, compared with 51% for GPT-6 Astra (max) and 54% for GPT-6.1 Sol (max). This means Argon is much more likely to acknowledge when it does not know an answer rather than guess incorrectly. On accuracy, Gemini 4 Argon scores 50%, a 5 point decrease from Gemini 3.1 Pro Preview, and 13 points below GPT-6 Astra (max, 63%). With this slightly lower accuracy, its overall AA-Omniscience score of 42 remains in line with GPT-6 Astra (43) and GPT-6.1 Sol (42)

Key model details:

➤ Context Window: 1M tokens

➤ Multimodality: Text, image, video, and speech input, with text output

➤ Pricing: $4/$20 per 1M input/output tokens at standard pricing, currently discounted 50% to $2/$10. Cached input tokens receive a 95% discount ($0.10 per 1M at discounted pricing), up from 90% on Gemini 3.8 Flash

➤ Long Decode Continuation: We tested Gemini 4 Argon with Long Decode Continuation, a new Gemini API feature that pauses long responses and resumes them across follow-up calls. This lets reasoning run up to 1M output tokens without request timeouts

[Shubham Saboo @Saboo_Shubham_]

INSANE. Voice AI Agents getting better every other week.

I'm now talking to a live avatar of my insurance claim voice agent. It can see, talk, think and draw in real-time now with a face.

Switched to Hindi mid-call and it still worked.

Made it 100% open-source.

[W. Ronny Huang @wronnyhuang]

it's a very general model. upload your own avatar image, voice sample, and character description, and then chat with it live. 30s demo here on vertex. unfortunately, must have enterprise account, but regular accounts can still use stock avatars (2/N)
https://t.co/nPIuDPUq6O

[W. Ronny Huang @wronnyhuang]

https://t.co/4AlmTMObT3
now making live avatar available to all! my most fun project at google thus far. started as a bottom-up research project on video agents with a small nimble team in gemini audio, supported by private customers on cloud, to now this general release (1/N)

[W. Ronny Huang @wronnyhuang]

Actually instead of thread I'm just pasting the write up here

https://t.co/iZVQYH8L6T

[W. Ronny Huang @wronnyhuang]

I spent the past 24 hours setting up open claw and playing with it

Thread

[W. Ronny Huang @wronnyhuang]

Culmination of a year's work by whole team. Lots of improvements https://t.co/50nE5ZvRGc

[W. Ronny Huang @wronnyhuang]

A year's worth of work by the entire team, and totally different underlying model/infra/method than before

[Josh Woodward @joshwoodward]

We just shipped our biggest update to Gemini Live ever. It's smarter. It's more expressive. It has accents. And my favorite: it can talk faster! Here are 3 ways people are already using it in @GeminiApp. 🧵

[W. Ronny Huang @wronnyhuang]

pushing the pareto front of reasoning performance and time to first audio is a worthy challenge and we'll keep iterating!

[Artificial Analysis @ArtificialAnlys]

Google’s Gemini 2.5 Native Audio Thinking is the new leading Speech to Speech model per our Artificial Analysis Big Bench Audio benchmark

The new model achieves a score of 92% on Big Bench Audio, the highest result recorded by Artificial Analysis to date. This not only places it ahead of all previously tested native Speech to Speech systems, but also above a GPT-4o pipeline approach (Whisper transcription → GPT-4o text reasoning → speech generation).

Benchmark context: Big Bench Audio is the first dedicated dataset for evaluating reasoning performance of speech models. Big Bench Audio comprises 1,000 audio questions adapted from the Big Bench Hard text test set, chosen for its rigorous testing of advanced reasoning, translated into the audio domain.

Performance:
➤ Reasoning: Achieves 92% on Big Bench Audio, setting a new state-of-the-art for native Speech to Speech reasoning
➤ Latency: At an average time to first token of 3.87 seconds, the new model is slower than leading OpenAI models including GPT Realtime (0.98 seconds), due to the thinking component. The non-thinking equivalent still leads on latency at 0.63 seconds

Model details:
➤ Processes audio, video, and text inputs directly, generating both text and natural speech outputs
➤ Reasons over spoken input without transcription
➤ Supports function calling, search grounding, and thinking budgets
➤ 128k input and 8k output token limits with a knowledge cut-off of January 2025

[W. Ronny Huang @wronnyhuang]

@thejetcoach Great to see alternative ways of traveling. If you guys launched a overnight route with sleeper beds between New York and Washington, I think that would be a game changer. It would save so much time.

[W. Ronny Huang @wronnyhuang]

@Delta I DMed @Delta my info

[W. Ronny Huang @wronnyhuang]

hey @Delta , we missed our flight b/c somehow your system booked my family's 4 tickets all under my name (that's not what I entered), and it took the gate agent over 40 minutes of waiting on hold to make a simple name change

[W. Ronny Huang @wronnyhuang]

They rebooked us onto the next flight which is 6 hours later, so we're now stuck in the airport with a 0 and 3 year old

[Tom Goldstein @tomgoldsteincs]

From the @icmlconf Chairs: "...we plan to reduce the number of accepted papers. Please work with your SAC to raise the bar. AC/SAC do not have to accept a paper only because there is nothing wrong in it."

[W. Ronny Huang @wronnyhuang]

combine with https://t.co/K7ovvyDSYY_latex() and the pain of transferring results to manuscript is gone https://t.co/BsODFaWuhi

[W. Ronny Huang @wronnyhuang]

awesome package for coloring tables in latex so ppl can read them more easily https://t.co/iT67LojVgw