Google DeepMind

Gemini 2.5 Pro

Type
language model
Context
1,049K tokens
Max output
66K
Released
17 June 2025
Knowledge cutoff
January 2025
API string
google/gemini-2.5-pro

Prices by provider

Provider Input, $ per 1M Output, $ per 1M In our data since
Poe $0.87 $7 31 Jul 2026
Jiekou.AI $1.125 $9 31 Jul 2026
Databricks databricks/databricks-gemini-2-5-pro $1.25 $10 31 Jul 2026
neon $1.25 $10 31 Jul 2026
frogbot $1.25 $10 31 Jul 2026
auriko $1.25 $10 31 Jul 2026
perplexity-agent $1.25 $10 31 Jul 2026
modelis $1.25 $10 2 Aug 2026
302.AI $1.25 $10 31 Jul 2026
AIHubMix $1.25 $10 31 Jul 2026
Abacus.AI $1.25 $10 31 Jul 2026
CrossModel $1.25 $10 31 Jul 2026
Databricks databricks-gemini-2-5-pro $1.25 $10 31 Jul 2026
DeepInfra $1.25 $10 31 Jul 2026
FastRouter $1.25 $10 31 Jul 2026
GitHub Copilot $1.25 $10 31 Jul 2026
Google DeepMind $1.25 $10 31 Jul 2026
Google Vertex AI $1.25 $10 31 Jul 2026
Helicone $1.25 $10 31 Jul 2026
Kilo Code $1.25 $10 31 Jul 2026
LLM Gateway $1.25 $10 31 Jul 2026
Merge Gateway $1.25 $10 31 Jul 2026
NEAR AI $1.25 $10 31 Jul 2026
OCI Generative AI $1.25 $10 31 Jul 2026
OpenRouter $1.25 $10 31 Jul 2026
Requesty $1.25 $10 31 Jul 2026
SAP AI Core $1.25 $10 31 Jul 2026
Vercel $1.25 $10 31 Jul 2026
ZenMux $1.25 $10 31 Jul 2026
Cortecs $1.654 $11.024 31 Jul 2026
vercel_ai_gateway $2.5 $10 31 Jul 2026
NanoGPT $2.5 $10 31 Jul 2026
OrcaRouter $2.5 $15 31 Jul 2026

Only the base tier and only per-token prices are shown. Batch, discounted and cached rates, as well as prices per image or per second of video, do not go into this table: they cannot stand in the same column as a price per million tokens.

A single provider appears several times if it sells this model under different identifiers and at different prices — for example, at different weight precision. The identifier is shown next to the name. Identical offers that arrived from two sources under different spellings are merged into one row.

The date in the last column is the day this price first entered our collection. The price may well be older: before that day we simply were not recording it. It has not changed since — otherwise a new row with a new date would stand in its place.

Measurement results

Task set Result Run conditions Measured by
Arena Score in French 1,470.66 1,456–1,485
Arena Score in Spanish 1,469.59 1,457–1,482
Arena Score in Russian 1,460.7 1,454–1,467
Arena Score, overall 1,457.52 1,455–1,460
Arena Score in English 1,457.33 1,454–1,461
Arena Score, hard prompts 1,454.97 1,452–1,458
Arena Score, expert questions 1,454.85 1,448–1,462
Arena Score, creative writing 1,454.49 1,449–1,460
Arena Score, programming 1,452.45 1,448–1,457
Arena Score, multi-turn dialogue 1,452.3 1,448–1,457
Arena Score, mathematics 1,450.58 1,444–1,458
Arena Score, long queries 1,449.35 1,445–1,453
Arena Score, instruction following 1,437.05 1,433–1,441
Arena Score, understanding diagrams 1,271.93 1,265–1,279
Arena Score, text recognition in images 1,269.27 1,264–1,274
Arena Score, working with images 1,261.67 1,257–1,266
Arena Score, web development 1,224.23 1,208–1,240
Mock AIME 2024–2025 — olympiad problems 84.71 % · with a tuned harness Epoch evaluations
Creative writing (Lech Mazur’s evaluation) 83.8 % lechmazur/writing Github repository
GPQA Diamond — graduate-level questions 80.39 % · with a tuned harness Epoch evaluations
SWE-bench Verified — fixing bugs in repositories 57.56 % · with a tuned harness Epoch evaluations
SimpleQA Verified — factual accuracy 56 % Epoch evaluations
WeirdML — unusual machine learning tasks 54.03 % thinking budget: 16K WeirdML Leaderboard
DeepResearch Bench — deep research 49.71 % DeepResearchBench Leaderboard
ARC-AGI — generalising to unseen patterns 41 % thinking budget: 16K · with a tuned harness ARC Prize Leaderboard
Terminal-Bench — working in the command line 32.6 % · with a tuned harness Terminal-Bench v2 Leaderboard
FrontierMath, levels 1–3 24.56 % Epoch evaluations
GDPval — tasks from real occupations 23.3 %
Chess puzzles 15.83 % Epoch evaluations
APEX-Agents 6.6 %
ARC-AGI-2 4.86 % thinking budget: 32K
CritPt — physics problems 2 %
FrontierMath, level 4 — research-grade problems 0 % Epoch evaluations