Gemini 3.5 Pro
Google's 2M-token frontier model — the largest production context window, with a Deep Think reasoning mode.
About
Gemini 3.5 Pro is Google's flagship Gemini model, announced at I/O in May 2026 and reaching general availability in July after a rebuild — DeepMind reportedly scrapped the Gemini 2.5 Pro architecture mid-cycle over structural failures in recursive tool-calling and rebuilt from scratch. The headline spec is a genuine differentiator: a 2-million-token context window, double Flash's 1M and the largest of any production frontier model.
It ships with a 'Deep Think' reasoning layer for hard multi-step problems, but Google gates that capability behind the $250/month AI Ultra tier; the base model rolls out on the $20/month AI Pro tier. API pricing (~$15 in / $60 out per million tokens) was reported but not officially confirmed at launch, and Google released no official benchmark numbers at GA — so capability comparisons against Claude Opus 4.8 and GPT-5.5 rest on position, not published scores.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported: Google published no official benchmark numbers at general availability.
Independent: no verified third-party benchmarks; press positions it against Claude Opus 4.8 and GPT-5.5 with context length as the clear edge.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Existence, launch window, 2M-token context, and Deep Think gating confirmed across independent press and Google's I/O framing. Docked hard for: (1) Google shipped no official benchmarks at GA, so every capability claim is positioning; (2) API pricing is reported, not confirmed; (3) the most differentiating capability (Deep Think) is locked behind the $250/mo Ultra tier.