Gemini 2.5 Pro

Google's advanced thinking model for complex reasoning, coding, and long context.

Context window

1,048,576 tokens (1M) input; up to 65K output

Input / 1M tokens

$1.25

Output / 1M tokens

$10.00

Provider

Google DeepMind

Tiered by prompt size: input $1.25/1M and output $10.00/1M for prompts up to 200K tokens; input $2.50/1M and output $15.00/1M for prompts over 200K tokens (paid tier). · Data verified 2026-07-02

Gemini 2.5 Pro is Google's flagship reasoning model from the 2.5 generation, featuring built-in 'thinking' for stronger accuracy on complex tasks. It handles advanced reasoning, coding, mathematics, and scientific problems, and supports a 1M-token context window with multimodal input (text, image, audio, video). It reached general availability in mid-2025 and remains a top-tier option for demanding reasoning and long-context work, now sitting below the newer Gemini 3.x family.

Capability index

Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.

Coding

Reasoning

Math

Multimodal

Long context

Speed

Cost efficiency

How to access it

Available through Google AI Studio, the Gemini API, and Vertex AI. Create an API key at aistudio.google.com and call model 'gemini-2.5-pro'.

Get access →Documentation →

Strengths

✓Strong multi-step reasoning with built-in thinking
✓1M-token context window for large documents and codebases
✓Multimodal input across text, image, audio, and video
✓Solid coding and STEM performance

Best for developers who...

Complex reasoning and STEM problem-solvingLong-context document and codebase analysisHigh-quality multimodal understanding

When to choose it (and when not to)

Reach for Gemini 2.5 Pro when...

→Complex reasoning, analysis, or STEM tasks that benefit from a thinking model
→Processing very long inputs (long documents, large repos)
→Multimodal tasks needing high quality

Look elsewhere if...

✕Latency- or cost-sensitive high-volume workloads (use Gemini 2.5 Flash or a Flash-tier model)
✕You want the newest frontier agentic performance (consider Gemini 3.5 Flash / 3.1 Pro)
✕You need an open-weight or self-hosted model

How to use it

›Let the model think for hard problems; you can steer reasoning depth
›Keep prompts under 200K tokens where possible to stay in the cheaper pricing tier
›Provide structured instructions and examples for coding and formatting tasks
›Use system instructions to fix persona and output format

Quickstart

Python

from google import genai

client = genai.Client(api_key="YOUR_API_KEY")
response = client.models.generate_content(
    model="gemini-2.5-pro",
    contents="Explain the tradeoffs between quicksort and mergesort.",
)
print(response.text)

Install with `pip install google-genai`. Get an API key at aistudio.google.com.

API model id: gemini-2.5-pro