Google's latest Gemini 2.5 models are its biggest response to ChatGPT yet — and they're already live

Gemini, Google's family of AI models, are often seen as playing second fiddle to OpenAI's GPT family, which powers ChatGPT — but they've come a long way since a bumpy launch, with Google's AI proving itself to be a great fit for those looking for anything from a speedy research aide to a solid smartphone assistant.

Unlike ChatGPT, Gemini is designed to run across a multitude of services, be it drafting messages in Gmail, transcribing calls in Google Meet, or acting as a second set of eyes and ears as the AI assistant for future Android XR smart glasses.

Now, Google has announced it is rolling out its latest upgrades to the models that make Gemini so powerful, pushing preview models like Gemini 2.5 Pro and Flash out of preview and into general availability — and making a new Flash-Lite model open for testing.

Here's what's new, and, more importantly, what it means for you.

Gemini 2.5: What's new, minus the jargon

If you're new to Gemini, or you're just a casual AI user, then talk of models, tokens, and benchmarks is likely to cause more confusion than clarity. So let's keep it simple: Google now offers three updated Gemini 2.5 models designed to tackle different kinds of tasks; Flash, Pro, and a newly introduced Flash-Lite.

• Flash is your everyday option. Think of it like OpenAI's GPT-4o, the model that powers most ChatGPT interactions. It's good for chat, but it can also help with brainstorming, writing, and even translating. It's the all-around model that strikes a balance between brain power and speed.
  
• Pro is, as the name might imply, a much more in-depth model, suited to taking on more demanding, professional tasks like coding, problem solving, data analysis, or more complex, multi-step prompts. It's slower in output, but it is Google's most intelligent model to date.
  
• Flash-Lite is the newest model to get an upgrade, now available for testing in Google AI Studio and Vertex AI, and it's best suited for developers and enterprises looking for a fast way to clear high-volume workloads that don't require too much power to solve.

For those already familiar with Google's models, the new Gemini 2.5 family is better at memory, reasoning, and responses, with improved performance and accuracy across the board.

All three models support multimodal inputs (text, images, and audio), tool use (like code execution), and have a massive 1 million token context window — ideal for handling large documents or specific and detailed prompts.

Both Gemini 2.5 Flash and 2.5 Pro also feature improved native audio output for more natural-sounding conversations, and enhanced reasoning with Deep Think.

For developers, Gemini 2.5 Flash now features rebalanced pricing, with slightly more expensive input tokens being met with cheaper output tokens.

Gemini 2.5 Flash and 2.5 Pro can now be accessed freely on the Gemini web portal and the Gemini mobile app by selecting your model of choice from the drop-down menu at the top of the conversation window.

More from Laptop Mag

• My favorite AI tool just hit Google Search, and it's actually useful — try it yourself
• A simple mistake in the Meta AI app could expose your deepest secrets
• ChatGPT may be the smartest software ever, but this Pong-era game console can do something it can't