Google's Gemini 3.8 Flash trades lower per-token rates for higher total costs

3 min read
Source: The Verge
Google's Gemini 3.8 Flash trades lower per-token rates for higher total costs
Photo: The Verge
TL;DR

Google has released Gemini 3.8 Flash, a new AI model that performs more reasoning steps than its predecessor but may cost users more overall. The model maintains the same introductory per-token pricing as Gemini 3.7 Flash, but Artificial Analysis notes that total costs have risen by approximately 40% due to increased output tokens. Google also launched Gemini 3.8 Flash Cyber for government partners through its new Fairwind Program.

Key points

  • Gemini 3.8 Flash is priced at $0.75 per million input tokens and $3.75 per million output tokens, matching the introductory rate of Gemini 3.7 Flash.
  • Artificial Analysis reports that the total cost per task is roughly 40% higher than the previous model, driven by a 30% increase in output tokens and more agentic turns.
  • The model outperforms competitors on the DeepSWE v1.1 software engineering benchmark, as well as the Vals Finance Agent V2 and Harvey Legal Agent benchmarks.
  • Google introduced Gemini 3.8 Flash Cyber alongside the Fairwind Program, a security initiative limited to governments and 650 trusted partners, including CrowdStrike.
  • The new model includes safeguards against misuse in Chemical, Biological, Radiological, and Nuclear (CBRN) and cyber offense domains.

Background

This release follows a rapid cadence of updates to Google's Flash models, including Gemini 3.7 Flash in August 2026 and earlier variants in July. The current model continues Google's strategy of offering lower introductory prices for new AI releases, though total costs may vary based on usage intensity.

How outlets are covering it

The Verge highlights the trade-off between performance and cost, noting that while per-token rates remain low, the model's tendency to use more tokens could increase overall expenses. The Register frames the release as a competitive move to maintain Google's position in the AI market. Artificial Analysis provides quantitative data showing that the total cost per task has risen significantly despite unchanged per-token pricing. Aigora.ai CEO John Ennis compares the model's coding quality to Anthropic's Opus 5, suggesting it offers similar performance at a lower cost.

Why it matters

The release of Gemini 3.8 Flash signals a shift in AI model economics, where performance gains may come at the expense of higher total costs for users. The introduction of the Fairwind Program also indicates a growing focus on cybersecurity and government partnerships in the AI sector.

What to watch

Developers and enterprise users can now access Gemini 3.8 Flash through Google AI Pro or Ultra subscriptions. The Fairwind Program will likely expand its membership as more government and security partners join. Future updates may focus on optimizing token usage to balance performance and cost.

Share this article

Want the full story? Read the original reporting

Read on The Verge