Google
Google

Gemini 2.5 Pro

Release Date 2025-06-17(Knowledge cutoff 2025-01-31)1.0M IN · 66K OUTProprietary ModelAPI available

Gemini Pro is Google's reasoning-focused model for text analysis of long documents, code and multimodal material. It is the deeper-analysis sibling of Flash in its generation, with search and tool support that still requires configuration and result checks by the surrounding application. AIB places research in the middle of the pack, a current comparative position rather than a guaranteed success rate. Google Cloud lists a retirement date that has passed, favoring migration to a supported successor rather than a new deployment; the separate Gemini API has its own lifecycle. Compare response time and total token use with correction effort, retries and service terms on representative inputs, rather than choosing solely from a demonstration or a headline benchmark.

API cost

Official price
$1.25IN$10OUT
SourceGoogle official docs

Output Speed

Fastper 2,000 tokens
105tok/s
First Output 2.5stotal: 21.6s
SourceOpenRouterLast updated

Expert Performance Evaluation

144586.2%

More about Arena Text

9.7%$57454.1%8.8%11.4%

More about τ³-Banking

69.0%143823.3%60.8%50.6%142341.6%

More about Long Context Reasoning

146241.5%24.6%11437.0%93.0%60.1

More about Arena Text Factuality

0.0%146478628.5%26.5%33.3122780.1%46.3%57.6%54.0%3.9%14600.4%32.6%1.51×

More about Terminal-Bench 4.0

1445144848.7%144314388.3873.6%1430

More about Arena Writing, Literature & Language

22.5%2.0%143984.4%84.7%87.787.7%24.6%0.0%1472144796.7%14.1%4.2%41.0%4.9%82.4%34.8

More about Humanity's Last Exam

126369.6%44.7%68.0%

More about Arena Vision

Specialized fields

1414144389.5%145814591469147090.0%58.1%74.3%46.9%63.2%63.9%46.9%93.2%93.8%92.8%82.5%80.0%31.6%95.7%84.7%0.5%77.2%15.8%78.2%85.9%75.3%45.8%84.9%24.4%5.0%26.0%

More about Arena Korean

OddBench

20.0%28400

More about BullshitBench v2

See each evaluation’s detail page for its source and last updateData corrections & removal requests