Gemini 2.5 Pro
Gemini 2.5 Pro is Google DeepMind's flagship reasoning model, generally available June 17, 2025. Built on Sparse MoE architecture, supports up to 1M token context, text/audio/image/video input, and integrated thinking mode.
Technical specification
Modalities
Capabilities
17Reasoning★
Reasoning
Multi-step reasoning★
Reasoning
Long context★
Reasoning
Coding★
Coding
Function Calling
Planning
Structured output★
Structured gen.
Audio understanding
Audio
Image understanding★
Vision
Video Understanding
Other
Chart understanding
Vision
Diagram reasoning
Reasoning
OCR★
Vision
Multilingual★
Language
Planning★
Planning
Streaming output
Reasoning
Interleaved Multimodal Input
Reasoning
Multimodal understanding★
Multimodality
Applications
Two-tier pricing based on context length. Prompts ≤200K tokens: $1.25/MTok input, $10.00/MTok output (thinking tokens counted as output). Prompts >200K tokens: $2.50/MTok input, $15.00/MTok output. Context caching: $0.31/MTok (≤200K), $0.625/MTok (>200K), storage $4.50/MTok/h. Batch API: ~50% discount. Free tier available in Google AI Studio (data used for product training).
INPUT
$1.2500 / per 1M tokens
OUTPUT
$10.0000 / per 1M tokens
CACHE
$0.3100 / per 1M tokens
TOTAL
for 10K tokens
Output includes thinking tokens. Context caching: $0.31/MTok read, storage $4.50/MTok/h.
Higher rate applies for long contexts exceeding 200K tokens.
Batch API (~50% discount). Asynchronous processing.
Security and enterprise
Model evaluated for cybersecurity, CBRN, autonomy, and other risks in accordance with Google DeepMind's Responsible Scaling Policy. Detailed safety assessments are included in the technical report and model card. Advanced mitigations against indirect prompt injection have been implemented.
Technical information
The technical report includes full safety evaluations covering cybersecurity, CBRN, Machine Learning R&D, and Deceptive Alignment. A model card is available at modelcards.withgoogle.com. At Google I/O 2025, Google announced significant improvements to protection against indirect prompt injection attacks, describing Gemini 2.5 as the "most secure model family to date". Deep Think mode underwent additional safety evaluations before broad release. Training data was subject to safety filtering. The paid API tier does not use data for model training, unlike the free tier.