Google Launches Gemini 3.6 Flash

🤖 AI Tool Alert — July 23, 2026

Google Launches Gemini 3.6 Flash

Released July 21, 2026 — a drop-in replacement for Gemini 3.5 Flash.

What changed:

• Pricing: $1.50/M input, $7.50/M output (down from $9.00/M — 16.7% cut)


• 17% fewer output tokens needed vs 3.5 Flash → 30-40% lower effective cost on agentic workloads


• 304 tokens/sec throughput — faster than 3.5 Flash


• Computer Use: 78.4% → 83% on OSWorld benchmark


• Knowledge cutoff: Jan 2025 → March 2026


• API ID: gemini-3.6-flash (swap model string in your code)


• Also launched: Flash-Lite at $0.30/M input for bulk tasks


• Google started pretraining Gemini 4 (no date yet)

Why you should care:
Drop-in replacement means a one-line change gets you better results for less. The 17% token reduction combined with cheaper pricing means multi-step agent workflows (research, code review, automated testing) cost significantly less per run.

Concrete example: Automated support triage averaging ~1,200 output tokens/ticket drops to ~996 tokens at $7.50/M → 30.5% cost reduction per ticket.

🔗 Google Blog | 9to5Google | Ars Technica