# Gemini 2.5 Pro

> Google's first thinking-native flagship — reasons before it answers, with a 1M-token context window.

- Maker: Google
- Line: Gemini Pro
- Canonical page: https://ai-tldr.dev/models/gemini-2-5-pro/
- Updated: 2026-06-18

## In short

Gemini 2.5 Pro is Google DeepMind's flagship Gemini Pro model, introduced on March 25, 2025. It has a 1M-token context window. You can use it through Google AI Studio / Gemini API and Google Vertex AI.

## Facts

| Fact | Value | Source |
| --- | --- | --- |
| Context window | 1M (max output 64K) | [1] |
| Input price | $1.25 / 1M tokens (prompts ≤200K); $2.50 / 1M tokens (prompts >200K) (cached $0.125 / 1M tokens (prompts ≤200K); $0.25 / 1M tokens (prompts >200K), plus $4.50 / 1M tokens per hour storage) | [2] |
| Output price | $10.00 / 1M tokens (prompts ≤200K); $15.00 / 1M tokens (prompts >200K) | [2] |
| Licence | Proprietary · API only | [1] |
| Released | 25 March 2025 | [3] |
| Parameters | Not disclosed | [1] |
| Knowledge cutoff | January 2025 | [1] |

## Sources

1. [Gemini 2.5 Pro model card (PDF)](https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Pro-Model-Card.pdf)
2. [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
3. [Google — Gemini 2.5 launch announcement](https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/)

## Overview

Gemini 2.5 Pro is Google DeepMind's flagship Gemini Pro model, introduced on March 25, 2025. It was Google's first "thinking-native" model: rather than answering immediately, Gemini 2.5 Pro reasons through a problem step by step before responding, which Google credits for its sharp gains on math, science, and coding benchmarks. At launch it debuted at the top of the LMArena human-preference leaderboard.

Gemini 2.5 Pro is natively multimodal. A single prompt can mix text, images, audio, video, PDFs, and entire code repositories, and the model ships with a 1-million-token context window (Google said a 2M window would follow), making it well suited to reasoning over very long documents and large codebases in one pass. Its knowledge cutoff is January 2025.

It is a proprietary, closed-weights model served through the Gemini API in Google AI Studio and on Vertex AI, as well as in the Gemini app for Advanced users. Google later promoted Gemini 2.5 Pro to general availability and refreshed it with preview updates; it has since been succeeded by the Gemini 3 Pro line.

## Benchmarks

- GPQA Diamond (science, pass@1): 84% ([source](https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/))
- AIME 2025 (math, pass@1): 86.7% ([source](https://www.datacamp.com/blog/gemini-2-5-pro))
- Humanity's Last Exam (no tools): 18.8% ([source](https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/))
- SWE-bench Verified (custom agent): 63.8% ([source](https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/))
- MMMU (multimodal, pass@1): 81.7% ([source](https://www.datacamp.com/blog/gemini-2-5-pro))
- Aider Polyglot (code editing): 74% ([source](https://www.datacamp.com/blog/gemini-2-5-pro))
- LiveCodeBench v5: 70.4% ([source](https://www.datacamp.com/blog/gemini-2-5-pro))

## Pricing

- Input: $1.25 / 1M tokens (prompts ≤200K); $2.50 / 1M tokens (prompts >200K) per 1M tokens
- Cached input: $0.125 / 1M tokens (prompts ≤200K); $0.25 / 1M tokens (prompts >200K), plus $4.50 / 1M tokens per hour storage per 1M tokens
- Output: $10.00 / 1M tokens (prompts ≤200K); $15.00 / 1M tokens (prompts >200K) per 1M tokens

Free tier available in Google AI Studio. Paid-tier prices are tiered by prompt size (≤200K vs >200K tokens).

Source: https://ai.google.dev/gemini-api/docs/pricing

## Strengths

- Thinking-native reasoning that leads on math (AIME) and science (GPQA) benchmarks without costly test-time tricks like majority voting
- 1M-token context window for reasoning over long documents, transcripts, and full code repositories
- Native multimodality across text, images, audio, video, and PDF in a single prompt
- Strong agentic coding, scoring 63.8% on SWE-bench Verified with a custom agent setup
- Topped the LMArena human-preference leaderboard at launch

## Best for

- Complex multi-step reasoning in math, science, and analysis
- Agentic coding and large-codebase refactoring across many files
- Long-document and long-context tasks (research papers, legal docs, transcripts)
- Multimodal understanding combining text, images, audio, and video
- Building developer agents and tools via the Gemini API or Vertex AI

## How to access

- [Google AI Studio / Gemini API](https://ai.google.dev/gemini-api/docs/pricing) — model id `gemini-2.5-pro`
- [Google Vertex AI](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/2-5-pro) — model id `gemini-2.5-pro`

## Gemini Pro versions

- [Gemini 3.5 Pro](https://ai-tldr.dev/models/gemini-3-5-pro/) (19 May 2026)
- [Gemini 3.1 Pro](https://ai-tldr.dev/models/gemini-3-1-pro/) (19 February 2026)
- [Gemini 3 Pro](https://ai-tldr.dev/models/gemini-3-pro/) (18 November 2025)
- Gemini 2.5 Pro (25 March 2025) — this page
- [Gemini 2.0 Pro](https://ai-tldr.dev/models/gemini-2-0-pro/) (5 February 2025)
- [Gemini 1.5 Pro](https://ai-tldr.dev/models/gemini-1-5-pro/) (15 February 2024)
- [Gemini 1.0 Ultra](https://ai-tldr.dev/models/gemini-1-0-ultra/) (8 February 2024)
- [Gemini 1.0 Pro](https://ai-tldr.dev/models/gemini-1-0-pro/) (13 December 2023)

## FAQ

### What is Gemini 2.5 Pro?

Gemini 2.5 Pro is Google DeepMind's flagship Gemini Pro model, released March 25, 2025. It was Google's first "thinking-native" model, meaning it reasons through a problem step by step before answering. It is natively multimodal and has a 1-million-token context window.

### What is the context window of Gemini 2.5 Pro?

Gemini 2.5 Pro has a 1-million-token context window and can generate up to about 64K tokens in a single response. At launch Google said a 2-million-token window would follow.

### How much does Gemini 2.5 Pro cost?

On the paid tier of the Gemini API, input is $1.25 per 1M tokens for prompts up to 200K tokens ($2.50 above 200K), and output is $10.00 per 1M tokens up to 200K ($15.00 above 200K). A free tier is available in Google AI Studio. (Source: Gemini API pricing page.)

### Is Gemini 2.5 Pro open source?

No. Gemini 2.5 Pro is a proprietary, closed-weights model. It is available only through Google's services — the Gemini API in Google AI Studio, Vertex AI, and the Gemini app — not as downloadable weights.

## Links

- [Google — Gemini 2.5 launch announcement](https://blog.google/technology/google-deepmind/gemini-model-thinking-updates-march-2025/)
- [Google DeepMind — Gemini models](https://deepmind.google/models/gemini/)
- [Gemini 2.5 Pro model card (PDF)](https://storage.googleapis.com/deepmind-media/Model-Cards/Gemini-2-5-Pro-Model-Card.pdf)
- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini 2.5 Pro on Vertex AI](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/2-5-pro)
- [Gemini 2.5 technical report (arXiv)](https://arxiv.org/pdf/2507.06261)
- [DataCamp — Gemini 2.5 Pro benchmarks](https://www.datacamp.com/blog/gemini-2-5-pro)
- [Artificial Analysis — Gemini 2.5 Pro](https://artificialanalysis.ai/models/gemini-2-5-pro)

Source: AI/TLDR LLM registry, https://ai-tldr.dev/models/gemini-2-5-pro/