Gemini 3.5 Flash: agents and coding at Flash speed
Google DeepMind introduced Gemini 3.5 with 3.5 Flash first, reporting it beats Gemini 3.1 Pro on agentic and coding benchmarks (Terminal-Bench 2.1 76.2%, MCP Atlas 83.6%) at about 4× the output speed of other frontier models.
What happened
Google DeepMind introduced Gemini 3.5, starting with 3.5 Flash, available in the Gemini app, AI Mode in Search, Google Antigravity, the Gemini API and enterprise products. 3.5 Pro was said to be in internal use and due the next month. The page is dated 19 May 2026.
The technical details
From the post:
- described as Google’s strongest agentic and coding model yet
- reported to beat Gemini 3.1 Pro on:
- Terminal-Bench 2.1: 76.2%
- GDPval-AA: 1656 Elo
- MCP Atlas: 83.6%
- CharXiv Reasoning (multimodal understanding): 84.2%
- about 4× faster in output tokens per second than other frontier models
Why it matters
The framing has moved from “chat” to “action”: long-horizon agent tasks. An MCP-based benchmark in a launch post shows how quickly MCP (my September 2026 study post) became standard plumbing.
Sources
- Gemini 3.5: frontier intelligence with action — Google DeepMind
This is a summary written from the linked public sources. Figures and numbers are as reported there.