Gemini Omni 1.1 Flash gives developers way more control over generative video. - Extends scenes up to 40 seconds total, built in 10 second increments - Now reads up to 10 seconds of prior context instead of just the last second, so continuity actually holds together - Lets you set first and last frames so camera moves and transitions look planned instead of random - 360p drafts render up to 60% faster and cost a third of what standard 720p costs, so you can test ideas cheap before committing - Upscales finished clips to 1080p or full 4K - Can reference up to 3 seconds of existing footage to keep characters and style consistent across clips - Already live in Google AI Studio, the Gemini Enterprise Agent Platform, and Google Flow for Plus, Pro, and Ultra subscribers - Adobe Firefly, Figma Weave, and Runway have already built it into their products Anyone here messing with AI video in production yet? https://blog.google/innovation-and-ai/technology/developers-tools/build-with-gemini-omni-1-1-flash

Google just dropped Gemini 3.5 Transcribe, and it's their most accurate speech-to-text model yet, built to handle real conversations instead of just clean studio audio. - Word error rate down to 4.0% for streaming and 2.6% for pre-recorded audio, measured by Artificial Analysis - That's a big jump from their old model, Chirp 3. Time to final transcription is 70% faster - Handles up to 85 languages and picks up on accents and dialects automatically - Cleans up your speech as you talk. Cuts filler words like "um" and "ah" and fixes self-corrections like "let's meet Tuesday, no wait, Wednesday" - Can identify up to 3 speakers in recorded audio with timestamps, which is huge for meeting notes and call analytics - It can also call other Gemini models to do stuff like generate images or analyze files, all triggered by voice - Already live in things like Rambler on Android and the Gemini app on macOS, and rolling into Chrome soon - Developers can plug it into their own apps now through the Gemini API and Google AI Studio If you're building any kind of voice agent, meeting notes tool, or automation that starts with "listen to this audio," this is worth testing against whatever you're using now. I still use Deepgram Nova 3 or Flux for my voice agents. But I might start using Gemini 3.5 Transcribe if I like the results. Will ultimately come down to latency. https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5-transcribe

Hi All, I sold an AI phone answering assistant. Thanks for the recent video on Trillet.ai Mason! I am curious if others here have done the same. If so: What did, do you charge for an AI phone assistant? What tech stack do you deploy for the service? Thanks in advance.

Welcome to Automate What Academy! Inside, you've got automation, agents, voice AI, and Claude Code content, plus paid courses if you want to go deeper. ✅ Drop a quick intro in THIS THREAD below ✅ - Where are you from? - Favorite AI tool? - What do you need help with right away?

Z.ai dropped GLM-5.3 on August 14, and the interesting part is not the benchmarks, it's that they didn't retrain anything. Same base model as GLM-5.2, every single gain came from scaled post-training. Here's what stood out to me: - Same 743B base model as GLM-5.2, zero new pretraining - Terminal-Bench 3.0 jumped from 4.6 to 28.3, that's long-horizon agent work - DeepSWE v1.1 went from 46.2 to 66.9 - CyberGym hit 84.5%, edging past Mythos 5 at 83.8% and GPT-5.6 Sol at 83.6% - ExploitBench more than doubled, 24.4% to 54.4% - Z.ai says the cyber jump was unplanned, capability kept compounding as they scaled training - Weights are held back roughly two weeks for safety evaluation and hardening - Live now through the Z.ai API, GLM Coding Plan and ZCode https://z.ai/blog/glm-5.3