Skip to content
← Models
ClosedGoogle DeepMind/10 trustVerified
Gemini 2.5 Deep Think logo

Gemini 2.5 Deep Think

Google's multi-agent parallel-reasoning model, first unveiled at I/O 2025, released August 2025.

Visit official page →

About

Gemini 2.5 Deep Think is Google DeepMind's multi-agent reasoning system, first introduced at Google I/O in May 2025 and rolled out to Google AI Ultra subscribers on August 1, 2025. Instead of following a single reasoning path, it spawns multiple AI agents that explore and evaluate hypotheses in parallel, then synthesizes the best solution.

A full research variant achieved gold-medal level performance at the 2025 International Mathematical Olympiad, while the consumer version available to $250-per-month Ultra subscribers reported strong results including 34.8 percent on Humanity's Last Exam and 87.6 percent on LiveCodeBench V6 (vendor-reported). It integrates code execution and Google Search tooling for complex coding, math, and scientific work.

Specs

Context window

Pricing

Free quota

API access

Weights

Closed — API only

Released

August 1, 2025

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

advanced-reasoningmath-and-sciencecodingresearch-assistance

Trust score — 9/10

/10 trustVerified

Scored 9/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 114; Trending score 86/100 from researcher assessment.

Sources