AGENTIC-SYSTEMSJune 17, 20251 min read17 views

We’re expanding our Gemini 2.5 family of models

Google is making Gemini 2.5 Pro and Flash generally available while introducing Gemini 2.5 Flash-Lite in preview, their most cost-efficient and fastest model yet.

We’re expanding our Gemini 2.5 family of models

We designed Gemini 2.5 to be a family of hybrid reasoning models that provide amazing performance, while also being at the Pareto Frontier of cost and speed. Today, we’re taking the next step with our 2.5 Pro and Flash models by releasing them as stable and generally available. And we’re bringing you 2.5 Flash-Lite in preview — our most cost-efficient and fastest 2.5 model yet.

Making 2.5 Flash and 2.5 Pro generally available

Thanks to all of your feedback, today we’re releasing stable versions of 2.5 Flash and Pro, so you can build production applications with confidence. Developers like Spline and Rooms and organizations like Snap and SmartBear have already been using the latest versions in-production for the last few weeks.

Introducing Gemini 2.5 Flash-Lite

We’re also introducing a preview of the new Gemini 2.5 Flash-Lite, our most cost-efficient and fastest 2.5 model yet. You can start building with the preview version now, and we’re looking forward to your feedback.

2.5 Flash-Lite has all-around higher quality than 2.0 Flash-Lite on coding, math, science, reasoning and multimodal benchmarks. 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. It comes with the same capabilities that make Gemini 2.5 helpful, including the ability to turn thinking on at different budgets, connecting to tools like Google Search and code execution, multimodal input, and a 1 million-token context length.

See more details about our 2.5 family of models in the latest Gemini technical report.

The preview of Gemini 2.5 Flash-Lite is now available in Google AI Studio and Vertex AI, alongside the stable versions of 2.5 Flash and Pro. Both 2.5 Flash and Pro are also accessible in the Gemini app. We’ve also brought custom versions of 2.5 Flash-Lite and Flash to Search.

We can’t wait to see what you continue to build with Gemini 2.5.

Source: Google DeepMind Blog

More in this category

agentic-systems

Align AI to Dynamic Human-AI Workflows

Current AI alignment methods fall short by relying on static representations of human preferences. A new paper proposes a paradigm shift toward "interactive and complementary alignment," allowing AI and humans to co-evolve within dynamic workflows.

agentic-systems

MemoHarness: Agent Harnesses That Learn from Experience

Researchers introduce MemoHarness, an adaptive harness optimization framework that enables AI agents to learn from their own executions. By analyzing performance and storing insights in a dual-layer experience bank, it dynamically adapts to new tasks, outperforming static configurations.

agentic-systems

How Artificial Intelligence LLM Engines Shape the Global Conflict Information Environment

A new study reveals how AI-powered answer engines are highly vulnerable to hallucination and manipulation when reporting on global conflicts. The lack of robust digital records allows malicious actors to exploit these systems through Generative Engine Optimization (GEO), posing a new threat of AI-driven information warfare.

agentic-systems

AI Agents Do Not Fail Alone:The Context Fails First

A new study validates context-engineering quality as an independent leading indicator of AI agent reliability, introducing an open-source evaluation infrastructure called ProofAgent-Harness to predict behavioral outcomes before deployment.

agentic-systems

Interpretable Language Model for Closed-Loop Type 1 Diabetes Control

Researchers have developed LLM-T1D, a novel approach combining reinforcement learning and large language models to automate insulin delivery for Type 1 Diabetes. The system not only outperforms traditional methods but also explains its decisions in plain language, addressing the 'black-box' challenge in medical AI.

agentic-systems

Capability from Access Structure, Not Scale: Lower Bounds and Pre-Registered Tests for Hybrid Sequence Models

A new study challenges the assumption that scaling alone guarantees capability convergence in AI models. Instead, researchers demonstrate that true capability depends on 'access structure'—specifically, hybrid architectures combining compressive state channels with scalable verbatim-index channels.