# Gemini 3.5 Flash on the Cotera agent benchmark

> Newer Flash, almost double the bill of the model it replaced. We'd still pick the older one.

Source: https://cotera.co/benchmarks/models/gemini-3.5-flash

---

Google's Gemini 3.5 Flash — the December 2026 update to the Flash family. Pitched as smarter than Gemini 3 Flash; in our matrix it was slower on the same tasks and ~2x the cost.

**Score:** 5/5 benchmarks passed · **Cost across the matrix:** $1.41

## Strengths
- Five for five, like its older sibling. Schema compliance was perfect.
- Marketing benchmark was actually cheaper here than on Gemini 3 Flash ($0.118 vs $0.218). The newer model is better at deciding when to stop reading Reddit.

## Watch-outs
- 36 tool calls on Coding. Read that again. 3.5 Flash decided the Stripe webhook task was a research project and kept poking at the docs until $0.487 ticked over.
- CX was $0.623 — nearly 2x the previous-gen Flash. The model is more thorough; the rubric didn't ask for more thorough.
- Total spend was $1.41 — more than GPT-5.5 and within striking distance of Gemini 3 Pro Preview. Use the cheaper Flash unless you've measured a quality gap on your workload.