Google Unveils Stable Gemini 2.5 Models and Previews Ultra-Efficient Flash-Lite

Google's Gemini AI family updated with stable 2.5 Pro, super-efficient 2.5 Flash-Lite

Google has announced a big expansion of its Gemini AI model family today. After months of tweaking and tuning, the high-power Gemini 2.5 Pro is leaving preview and is ready for developers to build on. Meanwhile, Google is offering a peek at its upcoming high-efficiency model, known as Gemini 2.5 Pro Flash-Lite. Try as it might, Google can't get away from confusing model names. Google's AI aspirations have been looking up in 2025 with the debut of Gemini 2.5. These models showed a marked improvement over past versions, making Google more competitive with OpenAI and its popular GPT models. However, we've been inundated with previews and test builds as Google works toward general availability, which means a model is stable enough for long-term development work. The 2.5 Flash model left preview at I/O, but Gemini 2.5 Pro lagged behind. Today, Flash is hitting general availability with the 04-17 build. Gemini 2.5 Pro is leaving preview and also reaching general availability, and as predicted, the recently revamped 06-05 build is the winner. This version aimed to address some issues that popped up in the Google I/O build of 2.5 Pro, and it appears to have worked.

Google's Gemini 2.5 models are out of preview and available to all

It's only been a few years since OpenAI showcased how powerful AI tools can be with ChatGTP. The AI tool took the world by storm, forever altering how we think and interact with AI models. It was very clear at that point, that going forward, this would be the future. Of course, Google, MIcrosoft, and other large companies would also step up to the plate, looking to introduce AI tools of its own. Related Google Gemini: Everything you need to know about Google's next-gen multimodal AI Google Gemini is here, with a whole new approach to multimodal AI Posts Google's efforts would eventually yield Gemini, which we've seen evolve over time, offering different tiers of service. We're now getting news from Google, sharing that its Gemini 2.5 Flash and Pro models will be available to all as they hit a stable release (via 9to5Google). In addition, the brand is also debuting a new model in preview called Gemini 2.5 Flash-Lite. With so many options, it can all be a bit confusing Source: Google Now, we get it, this can all be a little complicated if you're someone that just wants to dabble with AI and get a handle on Google's tools. If you're someone that pays for the service, you probably won't get much use out of this change, since it is aimed more towards developers. As mentioned before, all developers will now have the ability to access Google's Gemini 2.5 Flash and Pro models. Google has also introduced a preview of its new Gemini 2.5 Flash-Lite model, which it claims is its fastest model yet. Developers can start working with this model in preview starting today. Now, this can all be a bit confusing, but luckily the brand has issued a chart that breaks down how each of its Gemini 2.5 models works, and what it costs when it comes to inputs and outputs. You can check out the chart which breaks everything down in the image above, or if you're someone that likes to go a bit more in depth, you can always read the huge breakdown of Google's models if you want to really do a deep dive. If not, you can always check out our quick explainer on what Gemini is and what tools it offers. For the most part, this is just one more step in a long journey for AI. For most of us, it will be interesting to see just how these tools get utilized and how helpful they can be when they are introduced to new apps and services. Google will no doubt showcase its AI features in its latest products, like the Pixel 10, which we could see release really soon.

Google launches Gemini 2.5 Pro, free app access continues

Following the first preview three months ago, Google is ready to launch Gemini 2.5 Pro and make the reasoning model generally available (GA) for consumers and developers. Gemini 2.5 Pro entered experimental testing in late March (03-25). Four days after debuting for paying subscribers and developers, Google surprisingly made it available for free users. It got a big coding upgrade in May (05-06) just before I/O, and one last update at the start of June (06-05). With the stable version today, there are "no changes from the 06-05 preview version." L-R: Latest, 06-05, 05-06, 03-25 benchmarks 2.5 Pro will lose the "preview" label in the Gemini app's model picker. It follows 2.5 Flash entering GA in the app at I/O last month. Google notes how Pro is for "Reasoning, math & code" prompts, while Flash is for "Fast all-around help." Free Gemini users will continue to get "limited access" to 2.5 Pro, with Google AI Pro subscribers getting 100 prompts per day, or "expanded access." Google AI Ultra offers the "highest access." 2.5 Flash is also now generally available and stable for developers. It's the same 05-20 model preview from I/O with updated pricing: Additionally, developers can now preview Gemini 2.5 Flash Lite for "high-volume, latency-sensitive tasks like translation and classification" where cost is a priority. Google touts "lower latency than 2.0 Flash-Lite and 2.0 Flash on a broad sample of prompts." Thinking is turned off by default, but can be enabled with set budgets. Available native tools include "Grounding with Google Search, Code Execution, and URL Context in addition to function calling." There's also multimodal input and the 1 million-token context length.

Google updates Gemini 2.5 LLM series with new entry-level model, pricing changes - SiliconANGLE

Google LLC today introduced a new large language model, Gemini 2.5 Flash-Lite, that can process prompts faster and more cost-efficiently than its predecessor. The algorithm is rolling out as part of a broader update to the company's flagship Gemini 2.5 LLM series. The two existing models in the lineup, Gemini 2.5 Flash and Gemini 2.5 Pro, have moved from preview to general availability. The latter algorithm also received several pricing changes. Gemini 2.5 made its original debut in March. The LLMs in the series are based on a mixture-of-experts architecture, which means that they each comprise multiple neural networks. When a user submits a prompt, Gemini 2.5 activates only one of the neural networks rather than all of them, which lowers hardware usage. The LLM series is the first that Google trained using its internally-developed TPUv5p AI chip. According to the company, the training processing involved multiple server clusters that each contained 8,960 TPUv5p chips. Google's researchers equipped the clusters with new software that can automatically mitigate some technical issues. Gemini 2.5 models are multimodal with support for up to one million tokens per prompt. Google describes the flagship algorithm in the series, Gemini 2.5 Pro, as its most capable LLM to date. During internal tests, it outperformed OpenAI's o3-mini across a range of math and coding benchmarks. Gemini 2.5 Flash, the model that moved into general availability today together with Gemini 2.5 Pro, trades off some performance for efficiency. It responds to prompts faster and incurs lower inference costs. Gemini 2.5 Flash-Lite, the new model that Google debuted today, is an even more efficient model that is positioned as the new entry-level model in the LLM series. "2.5 Flash Lite has all-around higher quality than 2.0 Flash-Lite on coding, math, science, reasoning and multimodal benchmarks," Tulsee Doshi, the senior director of product management for Gemini, detailed in a blog post. "It excels at high-volume, latency-sensitive tasks like translation and classification, with lower latency than 2.0 Flash-Lite and 2.0 Flash on a broad sample of prompts." Gemini 2.5 Flash-Lite is billed at a rate of $0.1 per 1 million input tokens when developers submit prompts that contain text, images or video. That's less than one tenth the cost of Gemini 2.5 Pro. The price per million tokens of output, in turn, is $0.4 compared with $10 for Gemini 2.5 Pro. Google is changing the pricing of its mid-range Gemini 2.5 Flash model as part of the update. The company will now charge $0.3 per million input tokens and $2.5 per 1 million output tokens compared with $.15 and $3.5, respectively, before. Additionally, there is no longer separate pricing for tokens that the model processes in "thinking mode." The mode allows the LLM to boost output quality by increasing the amount of time and compute resources that it uses to generate prompt responses.

Google introduces stable Gemini 2.5 Flash and Pro, previews Gemini 2.5 Flash-Lite

Google DeepMind released stable versions of Gemini 2.5 Flash and Pro, highlighting top benchmark performance. Now broadly available to developers, the models support advanced reasoning, multimodal tasks, and one million-token context. A new Flash-Lite preview offers the fastest, most cost-efficient model in the Gemini 2.5 series.Google DeepMind has introduced the stable versions of its Gemini 2.5 Flash and Gemini 2.5 Pro models, its chief executive Sundar Pichai confirmed on Tuesday. ET reported in March that Gemini 2.5 Pro is the first release in the 2.5 series, which is claimed to be the most advanced reasoning AI model with "state-of-the-art" ranking across various benchmarks. Gemini 2.5 Pro got the top position on the large language model benchmarking tool LMArena, and received a score of 18.8% on Humanity's Last Exam, which has been specifically designed to test knowledge and reasoning capabilities. After being available only for Google AI Studio and Gemini app advanced users, the AI models have now become useful for developers. The tech major revealed that organisations like Snap Inc and SmartBear have already experimented with these models, in addition to developers like Spline and Rooms. In April, Google launched Gemini 2.5 Flash capable of real-time conversations, summaries, and extended interactions. It was available for both Google AI Studio and Vertex AI, paving the way for developers and companies to build AI agents. Following this, the company announced Gemini 2.5 Flash-Lite in preview. It claims that it is the most cost-efficient and fastest model in the series yet, which is efficient in coding, math, science, reasoning and multimodal benchmarks. It also supports a context length of up to one million tokens, which refer to words or pieces of a word used by AI systems.

Google rolls out stable Gemini 2.5 Flash and Pro, previews 2.5 Flash-Lite model

On Tuesday, Google announced the expansion of its Gemini 2.5 AI lineup by officially releasing the 2.5 Pro and 2.5 Flash models as stable and widely accessible. The company also introduced a preview of Gemini 2.5 Flash-Lite, highlighting it as the quickest and most affordable model in the 2.5 series so far. Tulsee Doshi, Senior Director of Product Management for Gemini, credited user insights for refining the 2.5 Pro and Flash releases. Over recent weeks, organizations including Snap, SmartBear, Spline, and Rooms have already put these models into live use. Google rolled out the Flash-Lite preview, which surpasses its 2.0 predecessor in benchmarks covering programming, mathematical problem-solving, scientific analysis, logical reasoning, and handling multiple input formats. Designed to handle large volumes of data with minimal delay, Flash-Lite responds faster than the earlier 2.0 Flash and 2.0 Flash-Lite models, making it ideal for tasks like translation and text classification. It accommodates up to one million tokens, supports varied input types, hooks into tools like Google Search and code execution, and allows dynamic scaling of compute resources to match budget needs. You can access the Gemini 2.5 Flash-Lite preview and the stable 2.5 Pro and Flash models through Google AI Studio, Vertex AI, or the Gemini app, and Google has also embedded customized Flash and Flash-Lite versions within its Search platform.