Gemini 2.5 Flash Lite Preview (09-2025) is Google's lightweight multimodal model supporting text, image, file, audio, and video inputs across a 1 million token context window. Its intelligence index of 23.8 and agentic index of 31.1 are modest, but GPQA at 0.793 and MMLU Pro at 0.842 indicate strong general knowledge. LiveCodeBench at 0.713 is respectable for a flash-tier model.

For businesses, the combination of broad modality support, 1M context, tool use, and very low pricing ($0.10 input / $0.40 output) makes this model highly attractive for high-volume content processing, document analysis, and multimodal pipelines. It is not suited to frontier reasoning tasks, but for structured content, summarisation, and instruction-following at scale, the value proposition is strong.

This is a preview release, so production stability should be validated before full deployment. Teams running cost-sensitive, high-throughput workflows with diverse input types will find it one of the most efficient options available at this price point.

Assessed June 30, 2026

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?