The example prompt for Google's new Gemini Flash TTS text-to-speed model is a lot https://
simonwillison.net/2026/Apr/15/ge
mini-31-flash-tts/
…
GENERATIVE AI
-
Google Gemini Flash TTS Model Demonstrates Advanced Text-to-Speech
By
–
-
Physion-Eval: Benchmarking Physical Realism in AI-Generated Videos
By
–
How physically realistic are our AI-generated videos? Physion Labs, Stanford University, MIT, Harvard University, and Character AI introduce Physion-Eval. This new benchmark uses expert human reasoning to meticulously diagnose and explain physical realism failures in
-
OCR Models Struggle with Non-Latin Unicode Scripts Benchmark
By
–
GlotOCR Bench OCR Models Still Struggle Beyond a Handful of Unicode Scripts paper: https://
huggingface.co/papers/2604.12
978
… -
Selective GPT API Dialog Access: Understanding Availability Differences
By
–
Oh see, I did not know that, never encountered these dialogs in my GPT API access long before I started. Is that selective then?
-
Trusted Tester Shares AI Platform Experience and Publishing Window Challenge
By
–
okay 🙂 im in trusted testers for whatever thats worth. wanted to make it title story for ainews, but window is gone as a consequence of the gradual rollout. just showing my side of things.
-
Google Gemini 3.1 Flash TTS Preview Speech Generation Tool
By
–
You can try this directly in AI Studio: https://
aistudio.google.com/generate-speec
h?model=gemini-3.1-flash-tts-preview
… -
MSL Publishes Preparedness Reports for Advanced AI Models
By
–
As we develop more capable models at the frontier, MSL is committed to safety and preparedness for AI. To demonstrate this commitment, we will be publishing preparedness reports for our models, in line with our new Advanced AI Scaling Framework. See our Muse Spark report below:
-
Single Image Transforms Into Immersive AI World
By
–
Turns a single image into a world. https://t.co/LVVy3lgmgq
— Robert Scoble (@Scobleizer) 15 avril 2026Turns a single image into a world.
-
NVIDIA Nemotron 3 Super: 120B Open Model for Agentic Reasoning
By
–
Banger paper from NVIDIA. Agentic reasoning needs models that are not just capable, but efficient at long-context inference. The agent model layer is moving toward open, long-context, high-throughput architectures. This paper introduces Nemotron 3 Super, an open 120B parameter
-
AI Safety Concerns: Autonomy Risks in Advanced Language Models
By
–
Never k*ll yourself. Wait for the next release of Claude to do it for you.