gemini4.ai

Update log

Gemini 4 updates

Every change to the picture, newest first. Entries are never rewritten silently — a claim that turns out to be wrong gets a dated correction of its own.

  1. CONF

    Baseline review: no Gemini 4 model exists in any Google source

    Checked the API changelog, the Google Cloud model list and Google's blog. No Gemini 4 model card, model ID, price or benchmark exists anywhere.

  2. LEAK

    First Gemini 4 Pro checkpoint reported, October window attached

    A first internal checkpoint for the next Pro flagship surfaced, with reporting attaching an October 2026 public release and a September Flash-Lite refresh.

  3. LEAK

    Leaked image puts output ceiling at 256k tokens and adds a high-effort mode

    Specifications read off a leaked image: a 256,000-token output limit, a multi-million-token context window, and a selectable high-effort compute mode.

  4. CONF

    Gemini 3.8 Flash ships while the Pro line stays frozen at 3.1

    Google released Gemini 3.8 Flash and 3.8 Flash Cyber, calling it its most intelligent workhorse model yet. The Pro flagship has not moved since February.

Subscribe via RSS. Labelling rules are on themethodology page.