Google Ships Gemini 3.5 Pro With a 2M Token Context Window
Google DeepMind released Gemini 3.5 Pro this week, the direct successor to March's Gemini 2.5 Pro, shipping with a built-in coding emulator and a refreshed pricing tier aimed squarely at Anthropic's Claude Opus 4.1.
What's covered:
Gemini 3.5 Pro release, internal codename, and the new built-in coding emulator
Lineage from Gemini 2.5 Pro (March release) and the iterative refinements since
2,000,000-token input context window and 64,000-token output ceiling
Updated API pricing tiers and how they compare to Claude Opus 4.1
Competitive landscape six months after the market looked settled around Anthropic
Roadmap: Gemini 3.5 Flash distilled variant and the next two scheduled releases
Timestamps:
00:00 Gemini 3.5 Pro coding breakthrough
00:02 Evolution from Gemini 2.5 Pro
00:38 Context window and API pricing
04:58 Competitive landscape and benchmarks
08:51 Upcoming releases and roadmap
Sources referenced: Google DeepMind release notes, Anthropic's Claude Opus 4.1 model card, and public benchmark results from LMArena and SWE-bench Verified.