Google is gifting its best AI model yet, Gemini 2.5 Pro (Exp), to free users

What you need to know

• Google brought Gemini 2.5 Pro experimental to all users starting March 29.
• Free users can try out Gemini 2.5 Pro experimental with rate limits, with Gemini Advanced users getting expanded access.
• Gemini Advanced subscribers can also take advantage of a one-million-token context window.

Google announced Gemini 2.5 Pro experimental less than a week ago, and it's already rolling out to free users on the web, plus the Gemini apps for Android and iOS. When the company revealed the early version of Gemini 2.5 Pro on March 25, it was exclusive to Gemini Advanced subscribers. That exclusivity period didn't last long, as Google extended its best experimental model to free users by Saturday, March 29.

"Gemini 2.5 Pro is taking off," the company wrote in a post on X (formerly Twitter) on the Gemini app account. "The team is sprinting, TPUs are running hot, and we want to get our most intelligent model into more people’s hands asap. Which is why we decided to roll out Gemini 2.5 Pro (experimental) to all Gemini users, beginning today."

Gemini 2.5 Pro experimental became available to users on the web client immediately, with availability on mobile apps said to be coming "soon."

There are still crucial advantages for Gemini Advanced subscribers. Google says that rate limits will apply to free Gemini users, whole Gemini Advanced users "have expanded access and a significantly larger context window." For now, Gemini 2.5 Pro experimental has a one-million-token context window that will double in the future. These features are available as part of the Google One AI Premium subscription, which costs $20 per month.

Gemini 2.5 Pro launched at the top many of the benchmarks and leaderboards for large language models (LLMs), including LMArena, where it is still the top-ranking chatbot as of today. Google originally explained that it developed Gemini 2.5 Pro by "combining a significantly enhanced base model with improved post-training."

Presumably, Google wants to get Gemini 2.5 Pro experimental — its best AI model yet — in the hands of more users. While still an experimental model, Gemini 2.5 Pro has support for file uploads, extensions, and Canvas. It currently lacks a few features, such as Deep Research mode.

It's unclear whether the final version of Gemini 2.5 Pro will also be made available to free users in a limited fashion.