modelbenchmark.io

Gemini 2.5 Pro

Google · gemini-2-5-pro

Compare

Google's proven reasoning model for coding, math, and multimodal analysis

Specification

google listing

Context
1M
Max output
66K
Released
2025-06-17
Knowledge cutoff
2025-01
Retires
2026-10-20
Open weights
no
Input
text, image, audio, video, pdf
Output
text

Price

US dollars per million tokens · google list price

Input
$1.25
Output
$10.00
Cache read
$0.125
Cache write
$0.375

Batch and priority prices

US dollars per million tokens · per listing

PriceListing$/M
Batch inputvertex_ai-language-models$0.625
Batch outputvertex_ai-language-models$5.00
Priority inputvertex_ai-language-models$2.25
Priority inputgemini$1.25
Priority outputvertex_ai-language-models$18.00
Priority outputgemini$10.00

Batch input is 50% below the interactive input price at the same listing.

Long-context price tiers

the rate above each context threshold

AbovePrice$/M
200KCache read$0.25
200KCache write$0.25
200KInput$2.50
200KOutput$15.00

Rate limits

as listed

Requests per minute
2,000
Tokens per minute
800,000

Quality

8 benchmarks · 2 sources

Composite
57th
percentile of 334 scored models
Rank
145 / 334
±0.06 sd
Evidence
8 × 2
3 or more benchmarks, from 2 or more sources
Effort range
4.0
points between effort settings
coding41st1/3 bench
reasoning83rd2/4 bench
math42nd5/7 bench

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
GPQA diamond85.3 ±2.1defaultEpoch AI2025-11-16
OTIS Mock AIME 2024-202584.2 ±4.8defaultEpoch AI2025-11-16
Chess Puzzles20.0 ±4.0defaultEpoch AI2025-12-08
FrontierMath-v127.1 ±2.0default · 2 runsEpoch AI2025-11-24
SWE-Bench verified57.6 ±2.3defaultEpoch AI2026-02-13
53.6default · mini-SWE-agentSWE-bench
Sources and settings disagree by 4.0 points. Both figures stand.
FrontierMath-Tier-4-2025-07-012.1 ±0.0default · 2 runsEpoch AI2025-07-03
FrontierMath-Tiers-1-3-v224.6 ±2.6defaultEpoch AI2026-06-11
FrontierMath-Tier-4-v20.0 ±0.0defaultEpoch AI2026-06-11

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 38 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
googlevendor$1.25$10.00$0.1251M66K
poe$0.87$7.00$0.0871.1M66K
jiekou$1.125$9.001M66K
302ai$1.25$10.001M66K
abacus$1.25$10.00$0.1251M66K
aihubmix$1.25$10.00$0.1251M66K
auriko$1.25$10.00$0.1251M66K
crossmodel$1.25$10.00$0.125$1.251M66K
databricks$1.25$10.00$0.125$1.251M66K
deepinfra$1.25$10.001M1M
fastrouter$1.25$10.00$0.311M66K
frogbot$1.25$10.00$0.311M66K
gemini$1.25$10.00$0.1251M66K
google-vertex$1.25$10.00$0.1251M66K
helicone$1.25$10.00$0.3125$1.251M66K
impossibl$1.25$10.00$0.1251M66K
kilo$1.25$10.00$0.125$0.3751M66K
llmgateway$1.25$10.00$0.1251M66K
llmgateway-providers · google-ai-studio$1.25$10.00$0.1251M66K
llmgateway-providers · google-vertex$1.25$10.00$0.1251M66K
merge-gateway$1.25$10.00$0.1251M66K
modelis$1.25$10.001M66K
nano-gpt$1.25$10.00$0.125$0.3751M66K
nearai$1.25$10.00$0.1251M66K
oci · oci$1.25$10.001M66K
ofox$1.25$10.00$0.125$4.501M66K
openrouter$1.25$10.00$0.125$0.3751M
66K
66K
orcarouter$1.25$10.00$0.1251M66K
perplexity-agent$1.25$10.00$0.1251M66K
sap-ai-core$1.25$10.00$0.1251M66K
vercel$1.25$10.00$0.1251M66K
vertex_ai-language-models$1.25$10.00$0.1251M66K2026-10-20
zenmux$1.25$10.00$0.31$4.501M64K
cortecs$1.495$9.964$0.242$0.4341M66K
vercel_ai_gateway · vercel-ai-gateway$2.50$10.001M66K
anyapi1M66K
github_copilot · github-copilot128K64K
qiniu-ai1M66K

Price history

append-only observations · a listing writes a row only when its price moves

DateHostField$/M
2026-09-02openrouterCache read$0.125
2026-09-06openrouterCache read$0.125
2026-09-02openrouterCache write$0.375
2026-09-06openrouterCache write$0.375

More from Google

most-hosted first