Loading page
Google
Gemini 2.0 Flash-Lite
Gemini 2.0 Flash-Lite is a lightweight chat model from Google, positioned as the smaller, higher-throughput option within the Gemini 2.0 Flash line.
Input from
$0.037 / 1M tokens
across 1 provider
API Pricing
Cheapest on Google Cloud — 33% below avg| Provider | Input / 1M | Output / 1M |
|---|---|---|
| $0.037 | $0.150 | |
| $0.075 | $0.300 |
Prices updated daily. Last check: Sep 22, 2026
Gemini 2.0 Flash-Lite pricing by provider
Input, output, and batch rates, plus alternatives, for one provider at a time.
Performance & Benchmarks
Source: Artificial Analysis →Intelligence
7.4 / 100
Reasoning & Knowledge
- MMLU-Pro72.4%
- GPQA Diamond53.5%
- Humanity's Last Exam3.4%
Coding
- LiveCodeBench18.5%
Math
- AIME27.7%
- MATH-50087.3%
Benchmarks measured Sep 2026. Scores are independent evaluations, not vendor-reported.
Model Details
General
- Creator
- Modalities
- Text
Capabilities
- Tool Calling
- No
- Open Source
- No
Strengths & Limitations
Strengths
- Positioned as the lightest tier of the Gemini 2.0 Flash line, aimed at cost-sensitive, high-request-volume deployments
- Available through Google's first-party surfaces (Gemini API and Vertex AI), so it fits existing Google Cloud auth, quota, and billing setups
- Flash-tier design targets low response latency, which suits interactive and streaming UI use cases
- Sits alongside Gemini 2.0 Flash in the same API family, making it straightforward to escalate hard requests to the larger sibling without changing SDKs
- Suitable as a routing or pre-filter model in tiered inference architectures where most traffic never needs a larger model
- Listed by multiple providers in our catalog, allowing side-by-side cost comparison in the pricing table on this page
Limitations
- Lower capability ceiling than Gemini 2.0 Flash and Pro-tier Gemini models on complex reasoning and long multi-step tasks
- We do not have confirmed throughput or time-to-first-token measurements for this model — the Artificial Analysis figures in our database are unpopulated
- Context window and supported input modalities are not verified in our metadata, so check Google's model documentation before relying on long-context or multimodal behavior
- Served as a hosted API model rather than something you run yourself on rented GPUs, so you cannot tune the serving stack for cost
- Superseded within Google's lineup by newer Gemini generations, which may affect long-term availability and support
Key Features
•Chat/text generation model in Google's Gemini 2.0 generation
•Flash-Lite tier — the lighter-weight variant of the Gemini 2.0 Flash line
•Access via the Gemini API and Vertex AI
•Streaming responses for interactive applications
•Designed for high-volume, latency-sensitive request patterns
•Drop-in family compatibility with Gemini 2.0 Flash for tiered model routing
•Listed across multiple inference providers for price comparison
About Gemini 2.0 Flash-Lite
Gemini 2.0 Flash-Lite is a text-generation chat model developed by Google as part of the Gemini 2.0 generation. Within that generation, the Flash line targets latency-sensitive and high-volume workloads, and the Flash-Lite variant sits below standard Gemini 2.0 Flash as the lighter-weight tier — intended for applications where per-request cost and response speed matter more than maximum reasoning depth.
The model is served through Google's API surfaces (the Gemini API and Vertex AI) and, in some cases, through third-party aggregators that resell access. Because it belongs to the Flash family rather than the Pro line, it is generally chosen for tasks that are well-specified and repetitive rather than open-ended multi-step reasoning. We do not currently track a confirmed context window, modality list, or throughput benchmark for this entry in our database; the benchmark figures we hold from Artificial Analysis are unpopulated, so any latency or tokens-per-second expectations should be validated against the specific provider you plan to use.
In practice, Flash-Lite models are deployed behind product features that run at scale — classification, extraction, summarization, and routing layers — where a heavier model in the same family would be used only for escalated or ambiguous cases. Compared with Gemini 2.0 Flash, Flash-Lite trades some capability headroom for a lower operating cost profile; compared with Gemini 2.0 Pro-class models, the gap in complex reasoning is larger still.
Common Use Cases
Gemini 2.0 Flash-Lite is aimed at workloads that run constantly and need predictable, low per-request cost: intent and topic classification, sentiment tagging, structured field extraction from documents or emails, short summarization, content moderation triage, query rewriting for search and RAG pipelines, and first-pass drafting inside product UIs. It also works well as the cheap tier in a routing setup, where it handles the bulk of straightforward traffic and hands ambiguous or complex requests up to Gemini 2.0 Flash or a Pro-tier model. It is a weaker fit for long agentic chains, hard mathematical or scientific reasoning, and large-scale code generation, where a higher-tier model in the Gemini family or a dedicated reasoning model is the more appropriate choice.
Frequently Asked Questions
How much does Gemini 2.0 Flash-Lite cost?
Pricing depends on the provider and on the pricing type — input tokens, output tokens, cached input, and batch requests are typically billed at different rates, and resellers may price differently from Google's own API. Check the pricing table on this page for current per-provider rates rather than relying on any figure quoted elsewhere.
What is Gemini 2.0 Flash-Lite best used for?
High-volume, well-defined text tasks: classification, extraction, short summarization, query rewriting, moderation triage, and as the low-cost first tier in a model-routing pipeline. It is less suited to deep multi-step reasoning or complex code generation.
How does it differ from Gemini 2.0 Flash?
Both are in the Gemini 2.0 Flash line, but Flash-Lite is the lighter-weight variant positioned below standard Flash. Flash-Lite is chosen when cost per request and speed dominate; standard Flash offers more capability headroom for harder prompts. Many teams run both and route between them.
Does Gemini 2.0 Flash-Lite support image input or a long context window?
Our database does not carry verified modality or context-window fields for this entry, so we do not state either way here. Consult Google's official model documentation for the authoritative capability list before building against those features.
Can I self-host Gemini 2.0 Flash-Lite on my own GPUs?
Gemini models are offered as a hosted API service by Google rather than as downloadable weights. If your requirement is to run inference on rented or owned hardware, compare open-weight alternatives and see our GPU pricing pages for the compute side of that cost.