Google · Model release
Gemini 3.8 Flash: workhorse-class coding at a fifth of Opus cost
Google's third Flash release in six weeks is a coding and agentic workhorse — and it ties Claude Opus 5 at the top of DeepSWE 1.1 at roughly a fifth of the cost per task.
What changed
- Tied #1 on Datacurve's independent DeepSWE 1.1 board (73.8% pass@1 at $2.36/task vs Opus 5's 73.6% at $11.84) on the mirror date.
- 90.8% on Terminal-Bench 2.1 (up from 81.6% for 3.7 Flash) and 58.7 on the AA Intelligence Index.
- Intro pricing $0.75/$3.75 per MTok through 2026-12-31 (then $1.50/$7.50); a restricted Gemini 3.8 Flash Cyber ships alongside it via the Fairwind Program.
Best fit
- Agentic coding at scale
- Mid-difficulty engineering
- Cost-aware frontier workloads
What to be careful about
- Google says it can consume more tokens than 3.7 Flash — 3.7 Flash stays available for efficiency-first workloads.
- Intro rates double on 2027-01-01; model that into any annual commitment.
Dataset snapshot: Gemini 3.8 Flash is also represented in the checked comparison dataset. Prices and benchmark figures carry their own field dates.