AI News

Google's Nano Banana 2 Lite (Gemini 3.1 Flash-Lite Image) launched June 30, 2026: 4-second generations at $0.034 per 1K image. How it compares to Nano Banana 2 and Nano Banana Pro, and when each tier

Sakana AI's Fugu Ultra is not one model — it is a trained orchestrator that assembles teams of specialist models behind a single endpoint. Here is what launched, the 73.7% SWE-Bench Pro claim, and wha

Twenty-plus notable models in under two months, three with million-token context windows in a single fortnight. The dated, complete list of 2026's AI model releases so far — and what the pattern says

Tencent's Hy3 is a 295B-parameter mixture-of-experts reasoning model priced at twenty cents per million input tokens. Here is why the July 6 release is the budget model worth actually watching.

Meta's Muse Spark 1.1 takes text, images, video, audio, and PDF documents in a single model with a 1M-token context. Here is what the everything-input agent model actually does, what it costs, and whe

Moonshot AI just shipped Kimi K3, a 2.8 trillion parameter open-weight multimodal model with a 1M-token context window. Here is what it actually does well, what it costs, and how to try it without tou

Google's Gemini 3.5 Flash quietly became the default brain behind the Gemini app, AI Mode in Search, and half of Google's agent stack. It's fast, cheap, and beats its own Pro sibling on agentic benchm

Anthropic launched Claude Fable 5 as a new tier above Opus, briefly had it export-controlled by the US government, then opened it to the world on July 1. Here's what Fable 5 is, how it differs from My

Anthropic's Claude Sonnet 5 landed June 30 with a 1M-token context window by default, adaptive thinking, and performance the company says is close to Opus 4.8 — at a fraction of the price. Here's the

Grok 4.20 Multi-Agent Beta doesn't just generate better answers—it changes how the work gets done. Here's what the multi-agent approach actually looks like and who should care.

Mistral just dropped three models worth testing side by side. Here's how Small 4, Devstral 2, and Ministral 3 14B actually differ when you put them to work.