Gemini Flash and Flash-Lite Versions Compared: Specs, Prices and Retirement Dates Side by Side

Gemini Flash and Flash-Lite are Google's faster, lower-cost Gemini models, and Google released six new Flash or Flash-Lite versions between May and September 2026, each with its own price and retirement date. This page compares every Flash and Flash-Lite text model on Google's Gemini API and Vertex AI documentation as of 15 November 2026, from Gemini 2.5 Flash to Gemini 3.8 Flash. It covers model codes, release dates, token limits, knowledge cutoffs, prices, capabilities and retirement dates, using identical fields for every model. It is written for developers, architects and buyers who run high-volume, agentic or extraction workloads on Flash-class models and need to pick a version or plan a migration. Models are listed by release date, newest first, an objective order that implies no ranking.

Last checked: 2026-11-15.

This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider. It uses Google's public Gemini API documentation as of November 2026. Model details change, so confirm against Google's current documentation.

Key takeaways

• Two models are named for new projects. Google's models page says "For any new projects, use our latest models: 3.5 Flash-Lite or 3.8 Flash."
• Two Flash versions are already rerouted. On 8 October 2026 the Gemini API deprecated gemini-3.7-flash and gemini-3.5-flash and routed them to gemini-3.8-flash and gemini-3.6-flash respectively.
• Gemini 3.8 Flash and 3.6 Flash double in price on 1 January 2027. Standard input moves from $0.75 to $1.50 and output from $3.75 to $7.50 per million tokens.
• Every model with a published limit lists 1,048,576 input tokens and 65,536 output tokens. That applies from Gemini 2.5 Flash to Gemini 3.8 Flash.
• Vertex AI retirement dates differ by model. Gemini 3.6 Flash retires on Vertex AI on 19 November 2026, Gemini 3.7 Flash on 28 January 2027, and Gemini 3.8 Flash on 2 September 2027 or later.

What are Gemini Flash and Flash-Lite models?

Google publishes each Gemini generation in tiers, and Flash and Flash-Lite are the faster, lower-cost tiers below Pro. Google describes Gemini 3.8 Flash as its "most intelligent Flash model", built for long-horizon software engineering, autonomous agents and complex enterprise workflows with Flash-level speed and cost efficiency. It describes Gemini 3.5 Flash-Lite as a low-latency, cost-effective multimodal model for high-throughput work such as subagent tasks and document parsing, and Gemini 2.5 Flash-Lite as suited to classification, simple data extraction and low-latency apps.

Google uses stable model codes (gemini-3.8-flash), preview codes (gemini-3-flash-preview) and a "latest" alias that, in Google's words, "will get hot-swapped with every new release of a specific model variation". The changelog records gemini-flash-latest moving to gemini-3-flash-preview on 21 January 2026 and to gemini-3.5-flash on 19 May 2026. The only Gemini 4 model on the models page, Gemini 4 Argon (gemini-4-argon), is not labelled Flash, so they are covered on the Google Gemini Models hub. See LLM Naming and Versioning Explained for aliases and snapshots in general.

Every model at a glance

As of 2026-11-15. Status is the label on Google's Gemini API models page; release dates are from the changelog and the Vertex AI model versions page.

Already shut down: gemini-2.0-flash and gemini-2.0-flash-lite (1 June 2026), gemini-3.1-flash-lite-preview (25 May 2026) and the Gemini 1.5 Flash models. Specialised Flash-named models such as image, TTS, Live, native audio and gemini-3.8-flash-cyber are out of scope here.

When was each model released?

As of 2026-11-15. Dates from the Gemini API changelog and the Vertex AI model versions page.

For every Gemini launch across 2026, see the LLM Release Tracker.

How much does each model cost?

Standard paid-tier prices in USD per million tokens, as of 2026-11-15. Source: the Gemini API pricing page on aistudio.google.com.

The pricing page lists cache storage per hour rather than a per-token cache write price, so the cache write column shows not stated; storage prices are in the notes. Output prices include thinking tokens, as the pricing page states. Batch rates are not available on the free tier. The doubling on 1 January 2027 applies to input, output, caching and storage for Gemini 3.8 Flash and 3.6 Flash; the pricing page lists no dated change for the other models. For a workload moving from Gemini 2.5 Flash to Gemini 3.8 Flash, standard input rises from $0.30 to $0.75 now and $1.50 from January 2027. For how caching, storage and thinking tokens affect a bill, see Model Pricing Explained and the LLM Pricing Tracker.

What changed between models?

Only changes Google documents on its model pages, pricing page, changelog and Vertex AI model versions page are listed.

• Gemini 2.5 Flash to 2.5 Flash-Lite: both list 1,048,576 input tokens, 65,536 output tokens and a January 2025 knowledge cutoff; Gemini 2.5 Flash-Lite adds PDF to the listed input types and lists lower prices ($0.10 input and $0.40 output against $0.30 and $2.50).
• Gemini 2.5 to Gemini 3 Flash: gemini-3-flash-preview launched on 2025-12-17, computer use support was added on 2026-01-29, and the changelog records gemini-flash-latest moving to gemini-3-flash-preview on 21 January 2026.
• Gemini 3 Flash to 3.1 Flash-Lite and 3.5 Flash: gemini-3.1-flash-lite became generally available on 2026-05-07 with thinking_level support and computer use listed as not supported; gemini-3.5-flash became generally available on 2026-05-19 and replaced gemini-3-flash-preview behind gemini-flash-latest. The changelog records computer use for Gemini 3.5 Flash in public preview from 2026-06-24.
• 3.5 generation to Gemini 3.6 Flash and 3.5 Flash-Lite: the changelog records temperature, top_p and top_k as deprecated with the 2026-07-21 release, and the pricing page lists one price for all input types for these models.
• Gemini 3.7 Flash and 3.8 Flash: the changelog records agentic video understanding for Gemini 3.7 Flash, 3.6 Flash and 3.5 Flash-Lite from 1 September 2026. The Gemini 3.8 Flash page lists thinking levels low, medium and high, and states that minimal returns an error. On 8 October 2026 the Gemini API deprecated gemini-3.7-flash and gemini-3.5-flash and routed them to gemini-3.8-flash and gemini-3.6-flash.
• Prices from 1 January 2027: the pricing page lists Gemini 3.8 Flash and 3.6 Flash input, output, caching and storage prices doubling on that date.

What workloads does Google position each model for?

As of 2026-11-15. This section records the provider's own positioning wording from the model pages and the Gemini API models page; it is not an assessment.

Google's models page also states: "For any new projects, use our latest models: 3.5 Flash-Lite or 3.8 Flash."

What is the context window of each model?

As of 2026-11-15, from each model's page on ai.google.dev or aistudio.google.com.

All of these models produce text only; Google lists audio generation, image generation and the Live API as not supported on each model page checked. The changelog records that temperature, top_p and top_k were deprecated with the 21 July 2026 release of Gemini 3.6 Flash and 3.5 Flash-Lite, so code that tunes sampling parameters should be reviewed when moving to a 3.6-generation or later model. See Context Windows Explained and Knowledge Cutoff Explained.

Capabilities and tools compared

As of 2026-11-15, from the capabilities list on each model page.

Every model page checked lists caching, code execution, file search, Grounding with Google Maps, Search grounding and URL context as supported, and Batch API, Flex inference and Priority inference as consumption options. The Vertex AI IDs come from the Vertex AI model versions page. The changelog records agentic video understanding for Gemini 3.7 Flash, 3.6 Flash and 3.5 Flash-Lite from 1 September 2026.

Where is each model available?

As of 2026-11-15, from the Gemini API models page, changelog and the Vertex AI model versions page.

• Gemini API and Google AI Studio: every current model on this page, with a free tier and a paid tier on the pricing page.
• Vertex AI: the model versions page lists gemini-3.8-flash, 3.7 Flash, 3.6 Flash, 3.5 Flash, 3.5 Flash-Lite, 3.1 Flash-Lite, 2.5 Flash and 2.5 Flash-Lite with release and retirement dates.
• Consumption options: the model pages list Batch API, Flex inference and Priority inference for each current model.

Model cards and system cards

As of 2026-11-15. Model card links are from the Google DeepMind model cards index; the Gemini 2.5 Flash card also covers Gemini 2.5 Flash Image. The index lists no separate system card or technical report for these models. This page does not summarise the cards' contents.

Benchmarks

The Gemini API model pages checked for these models give capability lists and descriptions rather than benchmark scores. Provider-reported benchmark figures for these models are not stated by Google on the pages checked (2026-11-15), and this page does not reproduce independent leaderboard figures. See How to Read LLM Benchmarks before comparing scores from other sources.

Deprecations and retirements

As of 2026-11-15, from the Gemini API changelog and models page and the Vertex AI model versions page.

The Gemini API and Vertex AI publish different lifecycles for the same model. On the Gemini API, Google states that the 2.5 models "are not deprecated and will continue to be served until further notice", for past users only, while Vertex AI lists their retirement as 20 October 2026. The Vertex AI page states that GA retirement dates "won't be moved to an earlier date than what is listed", and that once a retirement is scheduled for a shorter-availability model it gives at least 45 days to migrate. The Gemini 3 Flash preview has no retirement date on the pages checked. Across providers, see the LLM Deprecation and Retirement Tracker.

Data handling and security

• Unpaid Services: the Gemini API Additional Terms state that Google may use submitted content and responses to provide, improve and develop its products and machine learning technologies, that human reviewers may read inputs and outputs after they are disconnected from the account, and that users should not submit sensitive, confidential or personal information.
• Paid Services: the terms state that Google does not use prompts or responses to improve its products, and logs them for a limited time for policy enforcement, safety and legal requirements.
• EEA, Switzerland and UK: the terms apply the Paid Services data terms to all Services for users there, including Google AI Studio and unpaid quota.
• Grounding: the terms state that Grounding with Google Search stores prompts and output for thirty days.
• Keys and identities: see the Google Gemini API and AI Studio Security Guide, Workload Identity vs API Keys for LLM API Access and How to set spend limits on LLM API keys.

How to choose a model: factual criteria

For small models from other providers with the same fields, see Small and Low-Cost LLMs Compared and the sibling page OpenAI Mini and Nano Models Compared. For a single-model view, see Gemini 3.8 Flash.

Practitioner checklist

• Inventory every service calling gemini-3.6-flash on Vertex AI and migrate before 19 November 2026.
• Find callers of gemini-3.7-flash and gemini-3.5-flash on the Gemini API: they are already served by gemini-3.8-flash and gemini-3.6-flash, so re-run evaluations against the model actually answering.
• Pin stable model codes in production instead of gemini-flash-latest, which Google hot-swaps on each release.
• Budget for the 1 January 2027 price doubling on Gemini 3.8 Flash and 3.6 Flash, including caching and storage.
• Remove or review temperature, top_p and top_k settings when moving to 3.6-generation or later models.
• Keep sensitive or personal data off Unpaid Services outside the EEA, Switzerland and UK, as the terms advise.
• Track Vertex AI and Gemini API lifecycles separately, since dates differ for the same model.
• Scope and rotate the API keys used by high-volume callers, and set spend limits per key.

Standards and references

• Google: Gemini API models
• Google: Gemini API release notes (changelog)
• Google Cloud: Model versions and lifecycle
• Google: Gemini API pricing
• Google: Gemini 3.8 Flash model page
• Google: Gemini 3.5 Flash-Lite model page
• Google: Gemini 3.6 Flash model page
• Google: Gemini 3.1 Flash-Lite model page
• Google: Gemini 3 Flash Preview model page
• Google: Gemini 2.5 Flash-Lite model page
• Google: Gemini 2.5 Flash model page
• Google: Gemini API Additional Terms of Service
• Google DeepMind: Model cards

All sources accessed 2026-11-15.

Related NHI Mgmt Group resources: Google Gemini Models, Small and Low-Cost LLMs Compared, LLM Deprecation and Retirement Tracker, Model Pricing Explained.