> ## Documentation Index > Fetch the complete documentation index at: https://docs.apiyi.com/llms.txt > Use this file to discover all available pages before exploring further. # Gemini 3.5 Flash launches: Flash beats Gemini 3.1 Pro > Announced at Google I/O 2026, Gemini 3.5 Flash beats Gemini 3.1 Pro on Terminal-Bench 2.1 (76.2%), MCP Atlas (83.6%), Finance Agent v2 (57.9%) and GDPval-AA (1656 Elo). ~289 t/s output, around 4ร— faster than comparable frontier models. 1M context + 64K output + native multimodal. Listed at \$1.50/\$9.00 per 1M tokens (\$0.15 cached) โ€” identical to Google's official rates, up to 20% off via recharge bonus. **2026/5/20 10:04 (UTC+8)** ยท New Model ยท Google ๐Ÿš€ **Gemini 3.5 Flash launches ยท Flash beats Gemini 3.1 Pro** โ€” Google announced `gemini-3.5-flash` (no preview suffix) at **Google I/O 2026**. It beats Gemini 3.1 Pro on Terminal-Bench 2.1 (**76.2%**), MCP Atlas (**83.6%**), Finance Agent v2 (**57.9%**) and GDPval-AA (**1656 Elo**), with **\~289 t/s** output โ€” roughly **4ร— faster** than comparable frontier models. 1M context + 64K output + native multimodal (text / image / audio / video), with encrypted reasoning context preserved across API calls; inherits all Gemini 3 capabilities except Computer Use. ๐Ÿ’ฐ **Pricing matches Google's rates**: \$1.50 / \$9.00 per 1M tokens, \$0.15 cached input; non-global region \$1.65 / \$9.90. APIYi's direct-connection channel is live in sync โ€” stacking the [recharge promo](/en/faq/recharge-promotions) brings effective cost down to roughly **20% off** (\~\$1.20 / \$7.20 per 1M tokens). ๐Ÿ“– Launch notes: [/en/news/gemini-3-5-flash-launch](/en/news/gemini-3-5-flash-launch) *** โ† [Back to Live Updates](/en/live) ยท ๐Ÿ“š [Monthly Archive](/en/live/archive)