AI OrbitExplore • Compare • Stay Ahead
HomeAI ModelsGoogle DeepMindGemini 2.5 Pro
Gemini 2.5 Pro AI model logo
ActiveReleased Mar 25, 2025 Updated Sep 3, 2026

Gemini 2.5 Pro

By Google DeepMind

Compare with... Benchmarks
COMMUNITY RATING Rate Gemini 2.5 Pro Be the first to share a community rating.
Tap a star to rate instantly
Technical Performance composite69.4/100 verified composite
Context window1,048,576tokens
Input price$1.25per 1M tokens
Output price$10.00per 1M tokens
MODEL OVERVIEW

About Gemini 2.5 Pro

Gemini 2.5 Pro is a model from Google DeepMind, released on March 25, 2025. Its verified profile includes Text Generation, Reasoning, Image Understanding, Video Understanding, Audio Understanding, Multimodal, Code Generation, Structured Outputs, Function Calling, API Access, and Agent Workflows, with a 1,048,576-token context window.

Gemini 2.5 Pro is provided by Google DeepMind. AI Orbit records its release date as March 25, 2025. The current profile lists a 1,048,576 context window.

Cataloged capabilities include Audio Understanding, Text Generation, Reasoning, Image Understanding, Code Generation, and API Access. AI Orbit maps the model to use cases including General AI Assistance, Writing & Editing, Summarization, Brainstorming, and Question Answering.

The current AI Orbit benchmark composite is 69.4/100. 21 verified benchmark results are linked to this profile, including SWE-bench Verified, LOFT Hard <=128K, LOFT Hard 1M, and MRCR v2 8-Needle <=128K.

AI Orbit currently lists $1.25 per 1M input tokens and $10.00 per 1M output tokens. Provider pricing can change, so production decisions should be checked against the linked official pricing source.

Related model coverage includes Gemini 2.5 Flash, Gemini 3.7 Flash, Gemini 2.5 Flash-Lite, and GPT-5. AI Orbit currently links this model to 3 published comparisons. The profile also connects to current provider news when published intelligence is available.

ProviderGoogle DeepMind
ReleasedMar 25, 2025
Context window1,048,576
Benchmark composite69.4/100
Verified benchmarks21
Version—
Release dateMarch 25, 2025
StatusActive
Context window1,048,576
Associated product—
PROFILE EVIDENCE COVERAGE50/1005/8 checks verified

This score measures how much of the profile AI Orbit can support with stored evidence. It is not a rating of the model's quality, intelligence or performance.

IdentityExact version mapping still required
ProfileProfile verification is incomplete
Official evidenceOfficial-source evidence linked
CapabilitiesCapability taxonomy populated
Release dateOfficial release date stored
Context windowContext specification stored
Pricing modelCommercial terms are not fully pinned
BenchmarksVerified benchmark evidence available
How AI Orbit verifies model data
WHAT IT CAN DO

Capabilities & use cases

Audio UnderstandingAnalyzes speech, sound, music or other audio inputs.Text GenerationGenerates original text from natural-language instructions or structured prompts.ReasoningSupports multi-step reasoning, planning, analysis and problem solving.Image UnderstandingInterprets images, screenshots, diagrams and other visual inputs.Code GenerationGenerates source code, functions, components or complete implementations.API AccessProvides programmatic API access for application integration.MultimodalWorks across multiple modalities such as text, image, audio or video.Agent WorkflowsPlans and executes multi-step tasks with tools, state or autonomous actions.Video UnderstandingAnalyzes video content, scenes, speech and temporal context.Structured OutputsReturns schema-constrained JSON or other predictable structured data.Function CallingInvokes defined functions or tools with structured arguments.
Useful for
PERFORMANCE

Verified benchmark profile

AI Orbit only presents benchmark rows here when the underlying result is marked verified. Scores should still be interpreted alongside methodology and test date.

SWE-bench VerifiedUnclassified · Google Gemini 2.5 announcement · Mar 2025
63.8
LOFT Hard <=128KTechnical Performance · Google DeepMind
87.0
LOFT Hard 1MTechnical Performance · Google DeepMind
69.8
MRCR v2 8-Needle <=128KTechnical Performance · Google DeepMind
58.0
MRCR v2 8-Needle 1MTechnical Performance · Google DeepMind
16.4
MMMUTechnical Performance · Google DeepMind
82.0
ActivityNet-QATechnical Performance · Google DeepMind
66.7
EgoTempoTechnical Performance · Google DeepMind
44.3
Perception TestTechnical Performance · Google DeepMind
78.4
QVHighlightsTechnical Performance · Google DeepMind
75.0
VideoMMMUTechnical Performance · Google DeepMind
83.6
LiveCodeBenchTechnical Performance · Google DeepMind
74.2
Aider PolyglotTechnical Performance · Google DeepMind
82.2
SWE-bench Verified (Google Single Attempt)Technical Performance · Google DeepMind
59.6
SWE-bench Verified (Google Multiple Attempts)Technical Performance · Google DeepMind
67.2
GPQA DiamondTechnical Performance · Google DeepMind
86.4
Humanity's Last Exam (No Tools)Technical Performance · Google DeepMind
21.6
SimpleQATechnical Performance · Google DeepMind
54.0
FACTS GroundingTechnical Performance · Google DeepMind
87.8
Global MMLU LiteTechnical Performance · Google DeepMind
89.2
AIME 2025Technical Performance · Google DeepMind
88.0
API ECONOMICS

Pricing & availability

Commercial terms are shown in the unit that actually applies to this model instead of forcing every model into text-token pricing.

Source observed · verification pending
Pricing modelToken API
Billing unitNot classified
Evidence statusSource linked · review pending
Input tokensPer 1 million tokens$1.25
Output tokensPer 1 million tokens$10.00
2 token price source monitors · checked Aug 21, 2026 Official source
SOURCE TRANSPARENCY

Verification evidence

Official source coverage

AI Orbit separates structured evidence records, official provider links and benchmark-specific sources so missing evidence is not mistaken for a negative claim.

Official provider source linked

The profile is connected to an official provider page. Structured evidence records are shown separately when available.

Open source
Benchmark sources
COMMON QUESTIONS

Gemini 2.5 Pro FAQ

Quick answers based on the verified information currently available in this model profile.

What is Gemini 2.5 Pro?

Gemini 2.5 Pro is a model from Google DeepMind, released on March 25, 2025. Its verified profile includes Text Generation, Reasoning, Image Understanding, Video Understanding, Audio Understanding, Multimodal, Code Generation, Structured Outputs, Function Calling, API Access, and Agent Workflows, with a 1,048,576-token context window. The current AI Orbit pr...

Who created Gemini 2.5 Pro?

Gemini 2.5 Pro is provided by Google DeepMind.

What can Gemini 2.5 Pro do?

Its listed capabilities include Audio Understanding, Text Generation, Reasoning, Image Understanding, Code Generation.

What is Gemini 2.5 Pro useful for?

AI Orbit currently maps Gemini 2.5 Pro to use cases including General AI Assistance, Writing & Editing, Summarization, Brainstorming and Question Answering.

What is the context window of Gemini 2.5 Pro?

AI Orbit currently lists the context window as 1,048,576.

How much does Gemini 2.5 Pro cost?

AI Orbit currently lists $1.25 per 1M input tokens and $10.00 per 1M output tokens. Provider pricing can change, so check the linked official pricing source before production use.

GUIDES & ANALYSIS

Research related to Gemini 2.5 Pro

All AI guides

Published analysis is shown only when it connects through this model, its provider or its product category.

Benchmarks & EvaluationsAug 22, 2026How to Test an AI Model Before Production

Build a production-focused model evaluation using real tasks, acceptance thresholds, adversarial cases, latency, co...

Read analysis Benchmarks & EvaluationsAug 13, 2026AI Leaderboards: How to Read Rankings Without Being Misled

Read AI leaderboards critically by checking benchmark scope, score normalization, source methodology, model variant...

Read analysis Benchmarks & EvaluationsAug 12, 2026How to Compare AI Models Fairly

Create a fair model comparison using identical prompts, blinded scoring, known-answer tasks, latency measurement, c...

Read analysis
SIDE-BY-SIDE RESEARCH

Gemini 2.5 Pro comparisons

All model comparisons

Compare Gemini 2.5 Pro with other public AI models using linked specs, pricing and available benchmark evidence.

Gemini 2.5 Pro vs DeepSeek-R1Verified Aug 21, 2026Claude Sonnet 5 vs Gemini 2.5 ProVerified Aug 21, 2026GPT-5 vs Gemini 2.5 ProVerified Aug 21, 2026