Gemini 2.5 Flash is here
Gemini 2.5 Pro model is crazy good at coding, and now they have launched the new Gemini 2.5 Flash model (currently, in preview) which is crazy good at general tasks. Logan from Google describes it as:
Gemini 2.5 Flash is here, our first unified reasoning model with thinking budgets. 🔥
It’s on the perato frontier and punches above its price and size!!
Yes, it's a thinking model and is good with long contexts. Compared to OpenAI's new o4-mini, it's super cheap with almost comparable performance. It's priced at:
- Input price: $0.15/1M tokens
- Output price (no reasoning): $0.60/1M tokens
- Output price (with reasoning): $3.50/1M tokens
You can check and compare the pricing with other models on this website.
Gemini 2.5 Flash is already available at gemini.google.com and can also be used and tested via the AI Studio. And it's available via the API as well.
When testing through the API, I liked the fact that I can set the thinking_budget to control the reasoning effort it puts for the question.
Comment via email