# Gemini 2.5 Flash-Lite

> Google's cheapest, fastest Gemini 2.5 model for high-volume tasks

- Maker: Google
- Line: Gemini Flash-Lite
- Canonical page: https://ai-tldr.dev/models/gemini-2-5-flash-lite/
- Updated: 2026-06-18

## In short

Gemini 2.5 Flash-Lite is the smallest and cheapest model in Google's Gemini 2.5 family. Google released it on 17 June 2025 as an API-only model. Google first released it in preview on 17 June 2025 alongside the general availability of Gemini 2.5 Flash and Pro, then made Flash-Lite itself stable and generally available on 22 July 2025. It has a 1M-token context window.

## Facts

| Fact | Value | Source |
| --- | --- | --- |
| Context window | 1M (max output 64K) | [1] |
| Input price | $0.10 / 1M tokens (text / image / video) (cached $0.025 / 1M tokens (context cache, text / image / video)) | [2] |
| Output price | $0.40 / 1M tokens | [2] |
| Licence | Proprietary (Google), API-only · API only | [1] |
| Released | 17 June 2025 | [3] |
| Parameters | Undisclosed | [1] |
| Knowledge cutoff | January 2025 | [1] |

## Sources

1. [Gemini 2.5 Flash-Lite model docs](https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash-lite)
2. [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
3. [Flash-Lite is now stable and generally available](https://developers.googleblog.com/en/gemini-25-flash-lite-is-now-stable-and-generally-available/)

## Overview

Gemini 2.5 Flash-Lite is the smallest and cheapest model in Google's Gemini 2.5 family. Google first released it in preview on 17 June 2025 alongside the general availability of Gemini 2.5 Flash and Pro, then made Flash-Lite itself stable and generally available on 22 July 2025. It is positioned as a cost- and speed-optimized model for high-frequency, high-volume work such as classification, translation, data extraction, summarization, and intelligent routing.

Like the rest of the 2.5 line, Gemini 2.5 Flash-Lite is a 'thinking' model, but reasoning is off by default to keep latency and cost low. Developers can switch it on and set a controllable thinking budget when a task needs more careful reasoning, paying for the extra reasoning tokens only when they use them. The model accepts text, images, audio, video, and PDF input and returns text, with a 1,048,576-token (1M) context window and up to 65,536 output tokens.

It is Google's most affordable Gemini 2.5 model: $0.10 per million input tokens and $0.40 per million output tokens (audio input is priced higher at $0.30 per million). It also supports native tools including Grounding with Google Search, code execution, URL context, function calling, structured output, and context caching, and is available through both the Gemini API (Google AI Studio) and Vertex AI.

## Benchmarks

- GPQA diamond (science reasoning): 64.6% ([source](https://arxiv.org/abs/2507.06261))
- AIME 2025 (math): 49.8% ([source](https://arxiv.org/abs/2507.06261))
- MMMU (multimodal reasoning): 72.9% ([source](https://arxiv.org/abs/2507.06261))
- Global MMLU-Lite (multilingual): 81.1% ([source](https://arxiv.org/abs/2507.06261))
- FACTS Grounding: 84.1% ([source](https://arxiv.org/abs/2507.06261))
- Vibe-Eval (multimodal): 51.3% ([source](https://arxiv.org/abs/2507.06261))
- SWE-bench Verified (agentic coding): 31.6% ([source](https://arxiv.org/abs/2507.06261))
- Aider Polyglot (code editing): 26.7% ([source](https://arxiv.org/abs/2507.06261))
- SimpleQA (factuality): 10.7% ([source](https://arxiv.org/abs/2507.06261))
- Humanity's Last Exam: 5.1% ([source](https://arxiv.org/abs/2507.06261))

## Pricing

- Input: $0.10 / 1M tokens (text / image / video) per 1M tokens
- Cached input: $0.025 / 1M tokens (context cache, text / image / video) per 1M tokens
- Output: $0.40 / 1M tokens per 1M tokens

Audio input is $0.30 / 1M tokens. Batch and Flex tiers are 50% of standard rates; cache storage is $1.00 per 1M tokens per hour. Prices are for the paid tier.

Source: https://ai.google.dev/gemini-api/docs/pricing

## Strengths

- Lowest price in the Gemini 2.5 family ($0.10 input / $0.40 output per million tokens)
- Lowest latency in the 2.5 family, built for high-throughput, high-volume workloads
- Full 1M-token context window despite being the budget tier
- Multimodal input: text, images, audio, video, and PDF
- Optional thinking with a controllable, pay-only-when-used reasoning budget
- Native tools: Google Search grounding, code execution, URL context, function calling, structured output, and caching

## Best for

- High-volume classification and intelligent request routing
- Translation and large-scale summarization
- Data extraction from documents, PDFs, and images
- Latency-sensitive, cost-sensitive production pipelines
- Long-document and long-context processing on a budget
- Lightweight agentic tasks using search grounding and tool calls

## How to access

- [Google AI Studio / Gemini API](https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash-lite) — model id `gemini-2.5-flash-lite`
- [Google Vertex AI](https://cloud.google.com/vertex-ai/generative-ai/docs/models/gemini/2-5-flash-lite) — model id `gemini-2.5-flash-lite`
- [OpenRouter](https://openrouter.ai/google/gemini-2.5-flash-lite) — model id `google/gemini-2.5-flash-lite`

## Gemini Flash-Lite versions

- [Gemini 3.1 Flash-Lite](https://ai-tldr.dev/models/gemini-3-1-flash-lite/) (3 March 2026)
- Gemini 2.5 Flash-Lite (17 June 2025) — this page
- [Gemini 2.0 Flash-Lite](https://ai-tldr.dev/models/gemini-2-0-flash-lite/) (1 February 2025)
- [Gemini 1.5 Flash-8B](https://ai-tldr.dev/models/gemini-1-5-flash-8b/) (3 October 2024)

## FAQ

### When was Gemini 2.5 Flash-Lite released?

Google released Gemini 2.5 Flash-Lite in preview on 17 June 2025, alongside the general availability of Gemini 2.5 Flash and Pro. Flash-Lite itself reached stable general availability on 22 July 2025.

### How much does Gemini 2.5 Flash-Lite cost?

It is the cheapest Gemini 2.5 model: $0.10 per million input tokens (text, image, or video) and $0.40 per million output tokens. Audio input costs $0.30 per million tokens. Context caching and Batch/Flex tiers lower costs further.

### What context window and modalities does it support?

Gemini 2.5 Flash-Lite has a 1,048,576-token (1M) context window and produces up to 65,536 output tokens. It accepts text, images, audio, video, and PDF input and returns text output.

### Does Gemini 2.5 Flash-Lite support reasoning ('thinking')?

Yes. Like the rest of the Gemini 2.5 family it is a thinking model, but reasoning is turned off by default to keep it fast and cheap. Developers can enable it and set a controllable thinking budget, paying for reasoning tokens only when they use them.

## Links

- [Gemini 2.5 Flash-Lite model docs](https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash-lite)
- [Flash-Lite is now stable and generally available](https://developers.googleblog.com/en/gemini-25-flash-lite-is-now-stable-and-generally-available/)
- [Gemini 2.5: thinking model updates (June 17 2025 preview)](https://developers.googleblog.com/en/gemini-2-5-thinking-model-updates/)
- [Gemini API pricing](https://ai.google.dev/gemini-api/docs/pricing)
- [Gemini 2.5 technical report (arXiv 2507.06261)](https://arxiv.org/abs/2507.06261)
- [Gemini 2.5 Flash/Pro GA + Flash-Lite on Vertex AI](https://cloud.google.com/blog/products/ai-machine-learning/gemini-2-5-flash-lite-flash-pro-ga-vertex-ai)
- [OpenRouter: Gemini 2.5 Flash-Lite](https://openrouter.ai/google/gemini-2.5-flash-lite)

Source: AI/TLDR LLM registry, https://ai-tldr.dev/models/gemini-2-5-flash-lite/