17:I[87667,["/_next/static/chunks/546e34385bf0d4ef.js","/_next/static/chunks/8abeefe025a9cbcd.js","/_next/static/chunks/032bd8f98681d02a.js","/_next/static/chunks/f6ab42500d90b615.js","/_next/static/chunks/8debb43eacb9bc45.js","/_next/static/chunks/795a2b98b3bc1a3f.js","/_next/static/chunks/7ad92db2dbbaaec6.js","/_next/static/chunks/404df55c1cb97a6b.js","/_next/static/chunks/fc0d66f3910bc394.js","/_next/static/chunks/26262e36616117db.js"],"default"] 16:T529,{"@context":"https://schema.org","@graph":[{"@type":"CollectionPage","@id":"https://www.datalearner.com/en/ai-models/pretrained-models/Gemini-2_5-Flash-Lite/analysis#webpage","url":"https://www.datalearner.com/en/ai-models/pretrained-models/Gemini-2_5-Flash-Lite/analysis","headline":"Gemini 2.5 Flash-Lite Benchmark Results & Model Comparisons","description":"Gemini 2.5 Flash-Lite currently shows benchmark results led by MMLU (42 / 64, score 84.50), GPQA Diamond (109 / 162, score 66.70), AIME2025 (80 / 107, score 63.10).","author":{"@type":"Organization","name":"DataLearnerAI","url":"https://www.datalearner.com"},"publisher":{"@type":"Organization","name":"DataLearnerAI","logo":{"@type":"ImageObject","url":"https://www.datalearner.com/dl_logo_rect.webp"}},"mainEntity":{"@id":"https://www.datalearner.com/en/ai-models/pretrained-models/Gemini-2_5-Flash-Lite#software"},"mainEntityOfPage":{"@type":"WebPage","@id":"https://www.datalearner.com/en/ai-models/pretrained-models/Gemini-2_5-Flash-Lite/analysis#webpage"},"hasPart":[]},{"@type":"SoftwareApplication","@id":"https://www.datalearner.com/en/ai-models/pretrained-models/Gemini-2_5-Flash-Lite#software","name":"Gemini 2.5 Flash-Lite","url":"https://www.datalearner.com/en/ai-models/pretrained-models/Gemini-2_5-Flash-Lite","applicationCategory":"AI Model"}]}18:T653,

Gemini 2.5 Flash-Lite 是 Gemini 2.5 系列模型中最新发布的预览版本，以其高性价比和卓越的速度为特点。该模型在编码、数学、科学、推理和多模态基准测试中均展现出比 Gemini 2.0 Flash-Lite 更高的综合质量。特别是在处理大规模、对延迟敏感的任务，如翻译和分类时，Gemini 2.5 Flash-Lite 的表现尤为突出，其延迟低于 Gemini 2.0 Flash-Lite 和 Gemini 2.0 Flash。

Gemini 2.5 Flash-Lite 继承了 Gemini 2.5 系列的核心能力，包括灵活调整计算预算以实现不同程度的思考能力，支持连接 Google Search 等工具以及代码执行，并具备多模态输入和高达一百万 token 的上下文长度。

目前，Gemini 2.5 Flash-Lite 的预览版本已在 Google AI Studio 和 Vertex AI 上线，用户可以开始构建应用并提供反馈。同时，Gemini 2.5 Flash 和 Pro 的稳定版本也已正式发布，可供生产环境使用，并已集成到 Gemini 应用中。此外，Gemini 2.5 Flash-Lite 和 Flash 的定制版本也已应用于 Google Search。

Gemini 2.5 Flash-Lite Benchmark Details

Benchmark Results

Benchmark Results

综合评估