Model positioning
Released by Google on June 5, 2025, the upgraded preview of Gemini 2.5 Pro is a multimodal thinking model for complex reasoning and programming tasks. (blog.google)
Capabilities and uses
- Excels at code generation, complex programming, and agentic tasks, and can also be used for mathematics, science, and knowledge reasoning. (blog.google)
- Gemini 2.5 Pro supports text, image, audio, and video inputs, as well as a context window of approximately 1 million tokens; developers can control the reasoning process through a thinking budget. Use cases include long-document analysis, multimodal understanding, and developer tools. (storage.googleapis.com)
Limitations and status
- The model may produce incorrect or unreliable responses; code and important conclusions should be verified by humans.
- On June 17, 2025, Google announced that the general-availability version of Gemini 2.5 Pro was the same model as the 06-05 preview. The Gemini Developer API endpoint for this preview was shut down on December 2, 2025; as of November 18, 2026, the officially listed replacement model was
gemini-3.8-flash. (developers.googleblog.com)
Pricing
- Historical input and output prices for the 06-05 preview: unverified. The current official pricing page does not separately list prices for this discontinued preview, and current prices for other versions cannot be directly treated as its historical prices.
- General-availability version reference pricing, not representative of the preview's historical pricing: The Google Gemini Developer API standard paid tier for
gemini-2.5-prois billed in U.S. dollars (USD) per 1 million tokens. For prompts of no more than 200,000 tokens, input costs $1.25 and output costs $10; for prompts over 200,000 tokens, input costs $2.50 and output costs $15. Output pricing includes thinking tokens. Date checked: November 18, 2026. Official pricing page (aistudio.google.com)