Google released Gemini 3.5 Flash at I/O, scoring 55 on Artificial Analysis's Intelligence Index, within striking distance of rivals from Anthropic and OpenAI.
Gemini Flash Pricing
The model shipped into general availability across the Gemini API, Google AI Studio, Antigravity, Vertex AI, and AI Mode in Search. It is priced at $1.50 per million input tokens and $9.00 per million output tokens, roughly a third of what GPT-5.5 charges at $5.00 and $30.00. Anthropic charges $5 and $25 per million input and output tokens for Claude Opus 4.7, putting Google's new tier at roughly a third of that rate as well.
Independent benchmarker Artificial Analysis placed the model at 55 on its composite intelligence score, nine points above the older Gemini 3 Flash.
That puts it within two points of Claude Opus 4.7 and five points of GPT-5.5.
The release ships as the default model in the Gemini app and in AI Mode globally.
Also Read: Twenty One Capital Becomes Tether's Bitcoin Arm As SoftBank Walks Away
Pro Tier Looming
Google said a Pro version is in internal use and will arrive next month. If Flash already lands at 55, a Pro release would likely push into frontier territory currently occupied by Claude Opus 4.7 and GPT-5.5.
DeepMind chief technologist Koray Kavukcuoglu told reporters the Flash model "outperforms our latest frontier model, 3.1 Pro, on nearly all the benchmarks." Google reported scores of 76.2% on Terminal-Bench 2.1, 83.6% on MCP Atlas, and 1,656 Elo on GDPval-AA, all topping the February flagship.
Competitor Pressure
The launch arrives during user pushback against premium-priced flagships. Anthropic shipped Claude Opus 4.7 on Apr. 16, and within 48 hours developers documented token burn running 1.5 to 3 times higher than Opus 4.6 alongside reports of confidently broken output. OpenAI doubled its flagship API pricing with the GPT-5.5 release, drawing similar criticism from API customers.
Google's developer tools had steadily lost ground to Claude Code and OpenAI's Codex through late 2025 and early 2026, with Gemini 3.1 Pro receiving mixed community reception despite strong paper scores. Users reported uneven behavior inside third-party harnesses, citing a more clinical conversational tone in qualitative testing.
The Flash naming convention used to signal a cheaper, smaller model tier. With 3.5 Flash now outscoring the prior Pro flagship on coding and agentic suites, that distinction appears to be collapsing in Google's lineup.
Read Next: Security Experts Pour Cold Water On Claude Mythos Hacking Apocalypse
