
Google DiffusionGemma: Parallel LLM Hits 1,100 t/s
Google DeepMind open-sources DiffusionGemma, a 26B MoE model that generates 256 tokens per denoising pass instead of one at a time, reaching 1,100 tokens per second on a single H100.
They summarize our coverage. We write it.
Newsletters like this one rebroadcast our headlines - often without the full review, the source reading, or the analysis underneath. Our weekly briefing sends the work they paraphrase, straight from the desk, before they get to it.
Free, weekly, no spam. One email every Tuesday. Unsubscribe anytime.

Google DeepMind open-sources DiffusionGemma, a 26B MoE model that generates 256 tokens per denoising pass instead of one at a time, reaching 1,100 tokens per second on a single H100.

DiffusionGemma 26B is Google DeepMind's open-weight discrete diffusion language model that generates 256 tokens in parallel, reaching 1,100+ tokens/sec on H100 - roughly 4x faster than autoregressive models of the same size.

Google drops its entry AI subscription 38% to $4.99, undercutting every US rival and bringing the price war that reshaped India to American consumers.

Google DeepMind's new QAT checkpoints shrink the Gemma 4 E2B model to under 1GB, making serious on-device AI viable for phones and budget laptops.

Google will pay SpaceX $920 million per month for 110,000 NVIDIA GPUs at Colossus 1, citing unexpected demand for its Gemini Enterprise agent platform.

Alphabet priced an $84.75B equity raise on June 2, reversing a decade of share buybacks to fund a $180-190B AI infrastructure buildout - backed by a $10B bet from Berkshire Hathaway.

The UK's CMA issued a world-first binding order requiring Google to let publishers opt out of AI Overviews, AI Mode, and AI model training, with nine months to comply.

DuckDuckGo's no-AI search page saw a threefold traffic spike after Google's I/O 2026 overhaul made AI-generated summaries mandatory with no opt-out.

Google's Antigravity 2.0 rewrites the platform from a browser IDE into a five-surface agent suite. The architecture is ambitious, the launch was a mess.

Google is shutting down free access to its open-source Gemini CLI on June 18, replacing it with the proprietary Antigravity CLI - after accepting 6,000+ community pull requests.

YouTube now auto-detects and labels photorealistic AI video content without waiting for creator disclosure, using C2PA metadata and internal signals in a shift that removes creator control from the equation.

Gemini Spark is Google's first 24/7 cloud-persistent AI agent - ambitious, genuinely novel, and still rough around the privacy edges.