← Back to digest
Models & Releases

Gemini 3.8's price is introductory

What's this about?

Gemini 3.8 Flash costs the same as 3.7: $0.75 per million input tokens and $3.75 per million output. dotey's analysis flags the footnote — this is an introductory price that expires at the end of 2026, after which it doubles to $1.50 and $7.50.

Why does the footnote matter more than the headline?

Because people build cost models on today's price and deploy on a horizon longer than four months. A doubling is not a rounding error: any workload sized against $0.75 becomes a different business decision at $1.50, and by then you have written the integration. This is a standard tactic and it is disclosed — the problem is that the disclosure lives in a footnote while the number lives in the headline.

Why is it worth noticing?

  1. The expiry is roughly four months away, which is inside most procurement cycles.
  2. It sits alongside the week's other pricing subtleties: Anthropic's cheaper cache reads that do not change subscription metering, and the "20×" that is per-window.
  3. Every pricing story this week has the same shape — the headline number is true and describes something narrower than it appears to.
← Control-data flow separation in multi-agent LLMsOne model, two variants, refined by agentic loops →