For mobile and web apps, theFirebase AI Logic SDKs let you interactwith the supportedGemini models andImagen modelsdirectly from your app.

Gemini models are consideredmultimodal because they're capable ofprocessing and even generating multiple modalities, including text, code, PDFs,images, video, and audio.Imagen models can be prompted with text togenerate images.

Also, review ourFAQabout all the models thatFirebase AI Logic supports and does not support.

General use models

arrow_downward Jump to model comparisons

Image generating models

You can generate images with eitherGemini orImagen models.

arrow_downward Jump to model comparisons

Gemini

Imagen

Audio generating models

You can generatestreamed audio with models that support theGemini Live API.

arrow_downward Jump to model comparisons

The remainder of this page provides detailed information about the modelssupported byFirebase AI Logic.

• Compare models:

  • Supported input and output
  • High-level comparison of the supported capabilities
  • Specifications and limitations, for example max input tokens or max lengthof input video

• Description ofhow models are versioned, specifically theirstable,auto-updated,preview, andexperimental versions

• Lists ofavailable model names to include in yourcode during initialization

• Lists ofsupported languages for the models

At the bottom of this page, you canview detailed information about previous generation models.

---

Compare models

Each model has different capabilities to support various use cases. Note thateach of tables in this section describe each modelwhen used withFirebase AI Logic. Each model might have additionalcapabilities that aren't available when using our SDKs.

If you can't find the information you're looking for in the followingsub-sections, you can find even more information in your chosen API providerdocumentation:

• Gemini Developer API:Gemini models andImagen models

• Vertex AI Gemini API:Gemini models andImagen models

Note: We recommend reviewing details aboutthe location for where you access a model.TheGemini Developer API provides only global access to models, but theVertex AI Gemini API provides both global access (recommended for mostuse cases) and setting a specific location (supported locations depend on themodel).

Supported input and output

These are the supported input and output typeswhen using each model withFirebase AI Logic:

	Gemini 3 Pro, Flash	Gemini 3 Pro Image	Gemini 2.5 Pro, Flash, Flash-Lite	Gemini 2.5 Flash Image	Gemini 2.5 Flash- Live***	Imagen (generate)	Imagen (capability)
Input types
Text					(streaming)
Code
Documents (PDFs or plain-text)
Images
Video					(streaming)
Audio					(streaming)
Output types
Text
Text (streaming)					(transcription)
Code
Structured output (like JSON)
Images
Audio					(streaming)

To learn about supported file types, seeSupported input files and requirements.

Supported capabilities and features

These are the supported capabilities and featureswhen using each model withFirebase AI Logic:

	Gemini 3 Pro, Flash	Gemini 3 Pro Image	Gemini 2.5 Pro, Flash, Flash-Lite	Gemini 2.5 Flash Image	Gemini 2.5 Flash- Live***	Imagen (generate)	Imagen (capability)
Thinking
Generate text from text-only or multimodal inputs		interleaved or as part of image		interleaved or as part of image	transcription only
Generate images (Gemini orImagen)
Edit images (Gemini orImagen)
Generate audio					streaming only
Generate structured output (like JSON)
Analyze documents (PDFs or plain-text)
Analyze images (vision)
Analyze video (vision)					streaming only
Analyze audio					streaming only
Multi-turn chat					streaming only
Bidirectional multimodal streaming
System instructions
Count tokens

Supported tools

These are the supported toolswhen using each model withFirebase AI Logic:

	Gemini 3 Pro, Flash	Gemini 3 Pro Image	Gemini 2.5 Pro, Flash, Flash-Lite	Gemini 2.5 Flash Image	Gemini 2.5 Flash- Live***	Imagen (generate)	Imagen (capability)
Function calling
Code execution
URL context
Grounding with Google Search

Note:When usingFirebase AI Logic, the following capabilities arenot yet supported: context caching, fine tuning a model,embeddings generation, semantic retrieval, and someadvanced features ofImagen models.

Specifications and limitations

These are the specifications and limitationswhen using each model withFirebase AI Logic:

Property	Gemini 3 Pro, Flash	Gemini 3 Pro Image	Gemini 2.5 Pro, Flash, Flash-Lite	Gemini 2.5 Flash Image	Gemini 2.5 Flash- Live***	Imagen (generate)	Imagen (capability)
Input token limit*	1,048,576 tokens	65,536 tokens	1,048,576 tokens	32,768 tokens	32K (default; upgradable to 128K) tokens	480 tokens	480 tokens
Output token limit*	65,536 tokens	32,768 tokens	65,536 tokens	8,192 tokens	64K tokens	---	---
Knowledge cutoff date	January 2025	January 2025	January 2025	June 2025	January 2025	---	---
PDFs (per request)
Max number of input PDF files**	900 files	14 files	3,000 files	3 files	---	---	---
Max number of pages per input PDF file**	900 pages	14 pages	1,000 pages	3 pages	---	---	---
Max size per input PDF file	50 MB	50 MB	50 MB	50 MB	---	---	---
Images (per request)
Max number ofinput images	1,000 images	14 images	3,000 images	3 images	---	---	4 images
Max number ofoutput images	---	10 images	---	10 images	---	4 images	4 images
Max size per input base64-encoded image	7 MB	7 MB	7 MB	7 MB	---	---	---
Video (per request)
Max number of input video files	10 files	---	10 files	---	---	---	---
Max length of all input video (frames only)	~60 minutes	---	~60 minutes	---	---	---	---
Max length of all input video (frames+audio)	~45 minutes	---	~45 minutes	---	---	---	---
Audio (per request)
Max number ofinput audio files	1 file	---	1 file	---	---	---	---
Max number ofoutput audio files	---	---	---	---	---	---	---
Max length of allinput audio	~8.4 hours	---	~8.4 hours	---	---	---	---
Max length of alloutput audio	---	---	---	---	---	---	---

^{*For allGemini models, a token is equivalent to about 4 characters,so 100 tokens are about 60-80 English words. ForGemini models, you candetermine the total count of tokens in your requests usingcountTokens.}

^{**PDFs are treated as images, so a single page of a PDF is treated as one image. The number of pages allowed in a request is limited to the number of images the model can support.}

^{***Gemini 2.5 Flash-Live models are the native audio models that support theGemini Live API.}

Find additional detailed information

• Quotas andpricing aredifferent for each model. Pricing also depends on input and output.

• Learn about supported input file types, how to specify MIME type, and how tomake sure that your input files and multimodal requests meet the requirementsand follow best practices inSupported input files and requirements.

  Important:The total request size limit is 20 MB. To send large files, review theoptions for providing files in multimodal requests.

---

Model versioning and naming patterns

Models are offered instable,preview, andexperimental versions. Forconvenience, aliases without explicit version values are supported.

To find specific model names to use in your code, see the"available model names" section later on this page.

Important: StableGemini 2.5 model names donot have a three-digit suffix, and they donot have an auto-updated alias.

Version type / Release stage		Description	Model name pattern
Stable		Stable versions are available and supported for production use starting on the release date. A stable model version is typically released with a retirement date, which indicates the last day that the model is available. After this date, the model is no longer accessible or supported by Google.	Gemini 2.5 models Model names of stable versions have no suffix Example:gemini-2.5-pro Gemini 2.0 andImagen models Model names of stable versions are appended with a specific three-digit version number Example:gemini-2.0-flash-001 Example:imagen-3.0-generate-002
	Auto-updated stable alias (Gemini 2.0 models only)	Auto-updated stable aliases always point to thelateststable version of that model. If a new stable version is released, theauto-updated alias automatically starts pointing to that new stable version.	Gemini 2.0 models only Model names of aliases have no suffix Example:gemini-2.0-flash
Preview		Preview versions have new capabilities and are considerednot stable. These models arenot recommended for production use, come with more restrictive rate limits, and may have billing requirements. These models are retired within a few weeks or months after their associated stable version is released. For theVertex AI Gemini API, preview models released after June 2025 usually require you toset the model's location toglobal.	Model names of preview versions are appended with-preview and often the model's release date (-MM-DD for older models or-MM-YYYY for newer models) Examples: gemini-2.5-flash-preview-04-17 (released on April 17, 2025) or gemini-2.5-flash-preview-09-2025 (released in September 2025) or gemini-3-pro-preview (released in November 2025)
Experimental		Experimental versions have new capabilities and are considerednot stable. These models arenot recommended for production use and come with more restrictive rate limits. Experimental models are intended for gathering feedback and to enable experimentation with our latest features. These models are retired within a few weeks or months after their associated stable version is released. For theVertex AI Gemini API, experimental models released after June 2025 require you toset the model's location toglobal.	Model names of experimental versions are appended with-exp along with the model's release date (-MM-DD) Example:gemini-2.5-pro-exp-03-25 (released on March 25, 2025)
Retired		Retired versions are past their retirement date and have been permanently deactivated. Retired models are no longer accessible or supported by Google, and a request using a retired model name returns a 404 error.	---

For production use cases, we recommend using the explicit model name for the most recentstable version. Even though anauto-updated stable alias points to a stable version, the actual model version it points to will automatically change whenever a new stable version is released, which could mean unexpected behavior or responses.

Preview andexperimental versions are recommended duringprototyping only.

We also recommend usingFirebase Remote Config so that you candynamically change the model and version in your app without releasing a new version of your app.

---

Available model names

Model names are the explicit values that you includein your code duringinitialization of the model.

arrow_downward Jump toGemini model namesarrow_downward Jump toImagen model names

Programmatically list all available models

You can list all available models names using the REST API:

• Gemini Developer API: Call themodels.list endpoint

• Vertex AI Gemini API: Call thepublishers.models.list endpoint

Note that this returned list will includeall models supported by theAPI providers, butFirebase AI Logic only supports theGemini andImagen models described on this page.Also note that auto-updated aliases (for example,gemini-2.0-flash) aren'tlisted because they're a convenience alias for the base model.

Gemini model names

For initialization examples for your platform, see thegetting started guide.

For details about the release stages (especially for use cases, billing, andretirement), seemodel versioning and naming patterns.

Important: StableGemini 2.5 model names donot have a three-digit suffix, and they donot have an auto-updated alias.

Also, if you're using theVertex AI Gemini API, allGemini 3 andGemini 2.5preview models (exceptGemini Live API models) that are released after June 2025 areonly available in theglobal location.

Gemini 3 Pro model names

^{Requires the}pay-as-you-go Blaze pricing plan regardless of yourGemini API provider.

Model name	Description	Release stage	Release date	Retirement date
gemini-3-pro-preview	Preview version ofGemini 3 Pro	Preview	2025-11-18	To be determined

Gemini 3 Flash model names

^{Doesnot require the pay-as-you-go Blaze pricing plan if you're using theGemini Developer API (usually preview models require a paid plan).}

Model name	Description	Release stage	Release date	Retirement date
gemini-3-flash-preview	Preview version ofGemini 3 Flash	Preview	2025-12-17	To be determined

Gemini 3 Pro Image model names (aka "nano banana pro")

^{Requires thepay-as-you-go Blaze pricing plan regardless of yourGemini API provider.}

Model name	Description	Release stage	Release date	Retirement date
gemini-3-pro-image-preview	Preview version ofGemini 3 Pro Image (aka "nano banana pro")	Preview	2025-11-20	To be determined

Gemini 2.5 Pro model names

^{Doesnot require the pay-as-you-go Blaze pricing plan if you're using theGemini Developer API.}

Model name	Description	Release stage	Release date	Retirement date
gemini-2.5-pro	Stable version ofGemini 2.5 Pro	Stable	2025-06-17	No earlier than 2026-06-17

Gemini 2.5 Flash model names

^{Doesnot require the pay-as-you-go Blaze pricing plan if you're using theGemini Developer API.}

Model name	Description	Release stage	Release date	Retirement date
gemini-2.5-flash	Stable version ofGemini 2.5 Flash	Stable	2025-06-17	No earlier than 2026-06-17

Gemini 2.5 Flash‑Lite model names

^{Doesnot require the pay-as-you-go Blaze pricing plan if you're using theGemini Developer API.}

Model name	Description	Release stage	Release date	Retirement date
gemini-2.5-flash-lite	Stable version ofGemini 2.5 Flash‑Lite	Stable	2025-07-22	No earlier than 2026-07-22

Gemini 2.5 Flash Image model names (aka "nano banana")

^{Requires thepay-as-you-go Blaze pricing plan regardless of yourGemini API provider.}

Model name	Description	Release stage	Release date	Retirement date
gemini-2.5-flash-image	Stable version forGemini 2.5 Flash Image (aka "nano banana")	Stable	2025-10-02	No earlier than 2026-10-02

Gemini 2.5 Flash Live model names

^{Doesnot require the pay-as-you-go Blaze pricing plan if you're using theGemini Developer API (usually preview models require a paid plan).}

Gemini 2.5 Flash Live models are thenative audio models that supporttheGemini Live API. Even though the model has different model names dependingon theGemini API provider, the behavior and features of the model arethe same.

Gemini Developer API Model name	Description	Release stage	Release date	Retirement date
gemini-2.5-flash-native-audio-preview-12-20251	Latest preview version for theLive API on theGemini Developer API	Preview	2025-12-12	To be determined
gemini-2.5-flash-native-audio-preview-09-20251	Initial preview version for theLive API on theGemini Developer API	Preview	2025-09-18	To be determined

Vertex AI Gemini API Model name	Description	Release stage	Release date	Retirement date
gemini-live-2.5-flash-native-audio2	Stable version for theLive API on theVertex AI Gemini API	Stable	2025-12-12	No earlier than 2026-12-12
gemini-live-2.5-flash-preview-native-audio-09-20252	Preview version for theLive API on theVertex AI Gemini API	Preview	2025-09-18	To be determined

^{1Only supported by theGemini Developer API. Also, even though these are preview models, they're available on the "free tier" of theGemini Developer API.
2Only supported by theVertex AI Gemini API. Also, these models are not supported in theglobal location.}

Imagen model names

For initialization examples for your platform, see thegenerate images withImagen guide.

For details about the release stages (especially for use cases, billing, andretirement), seemodel versioning and naming patterns.

Important: AllImagen models require thepay-as-you-go Blaze pricing plan regardless of yourGemini API provider.

Imagen 4 model names

Model name	Description	Release stage	Release date	Retirement date
imagen-4.0-generate-001	Stable version ofImagen 4	Stable	2025-08-14	No earlier than 2026-08-14

Imagen 4 Fast model names

Model name	Description	Release stage	Release date	Retirement date
imagen-4.0-fast-generate-001	Stable version ofImagen 4 Fast	Stable	2025-08-14	No earlier than 2026-08-14

Imagen 4 Ultra model names

Model name	Description	Release stage	Release date	Retirement date
imagen-4.0-ultra-generate-001	Stable version ofImagen 4 Ultra	Stable	2025-08-14	No earlier than 2026-08-14

Imagen 3 Capability model names

Model name	Description	Release stage	Release date	Retirement date
imagen-3.0-capability-0012	Initial stable version ofImagen 3 Capability	Stable	2024-12-10	No earlier than 2025-12-10

^{2Not supported by theGemini Developer API no matter how you access it.}

---

Supported languages

Note: These languagesdo not represent the locations for accessing the model;instead, these are thelanguages that the models can understand and(forGemini) respond in (for example, the text input and output).If needed, seespecify the location for accessing a model.

Gemini

• All theGemini models can understand and respond in thefollowing languages:

  Arabic (ar), Bengali (bn), Bulgarian (bg),Chinese simplified and traditional (zh), Croatian (hr), Czech (cs),Danish (da), Dutch (nl), English (en), Estonian (et), Finnish (fi),French (fr), German (de), Greek (el), Hebrew (iw), Hindi (hi), Hungarian (hu),Indonesian (id), Italian (it), Japanese (ja), Korean (ko), Latvian (lv),Lithuanian (lt), Norwegian (no), Polish (pl), Portuguese (pt), Romanian (ro),Russian (ru), Serbian (sr), Slovak (sk), Slovenian (sl), Spanish (es),Swahili (sw), Swedish (sv), Thai (th), Turkish (tr), Ukrainian (uk),Vietnamese (vi)

• Gemini 2.0 Flash,Gemini 1.5 Pro andGemini 1.5 Flash models can understand and respond in thefollowingadditional languages:

  Afrikaans (af), Amharic (am), Assamese (as), Azerbaijani (az),Belarusian (be), Bosnian (bs), Catalan (ca), Cebuano (ceb), Corsican (co),Welsh (cy), Dhivehi (dv), Esperanto (eo), Basque (eu), Persian (fa),Filipino (Tagalog) (fil), Frisian (fy), Irish (ga), Scots Gaelic (gd),Galician (gl), Gujarati (gu), Hausa (ha), Hawaiian (haw), Hmong (hmn),Haitian Creole (ht), Armenian (hy), Igbo (ig), Icelandic (is), Javanese (jv),Georgian (ka), Kazakh (kk), Khmer (km), Kannada (kn), Krio (kri),Kurdish (ku), Kyrgyz (ky), Latin (la), Luxembourgish (lb), Lao (lo),Malagasy (mg), Maori (mi), Macedonian (mk), Malayalam (ml), Mongolian (mn),Meiteilon (Manipuri) (mni-Mtei), Marathi (mr), Malay (ms), Maltese (mt),Myanmar (Burmese) (my), Nepali (ne), Nyanja (Chichewa) (ny),Odia (Oriya) (or), Punjabi (pa), Pashto (ps), Sindhi (sd),Sinhala (Sinhalese) (si), Samoan (sm), Shona (sn), Somali (so), Albanian (sq),Sesotho (st), Sundanese (su), Tamil (ta), Telugu (te), Tajik (tg),Uyghur (ug), Urdu (ur), Uzbek (uz), Xhosa (xh), Yiddish (yi), Yoruba (yo),Zulu (zu)

Imagen

• General availability: English

• Preview: Chinese (simplified), Chinese (traditional), Hindi, Japanese,Korean, Portuguese, Spanish

---

Information about previous models

The following are active, but previous generation models. We recommend using oneof the latest models instead when possible.

Gemini 2.0 Flash andGemini 2.0 Flash‑Lite models will be retired on March 31, 2026 (stableGemini Live API 2.0 models are not impacted). AllGemini 1.0 models andGemini 1.5 are already retired, and all requests to these models return a 404 error.

To avoid service disruption, update to anewer model (for example,gemini-2.5-flash-lite).Learn more.

When you start seriously developing your app,we recommend usingFirebase Remote Config so that you canremotely change the model and version in your app without releasing a new version of your app.

Gemini models

• gemini-2.0-flash-001 (and its auto-updated aliasgemini-2.0-flash)
• gemini-2.0-flash-lite-001 (and its auto-updated aliasgemini-2.0-flash-lite)

For information about olderGemini Live API models, see theGemini API provider documentation:

• gemini-2.0-flash-live-001
• gemini-2.0-flash-live-preview-04-09
• gemini-live-2.5-flash-preview

Imagen models

• imagen-3.0-generate-002
• imagen-3.0-fast-generate-001

View supported input and output of previous generation models

These are the input and output typeswhen using each model withFirebase AI Logic:

	Gemini 2.0 Flash	Gemini 2.0 Flash- Lite
Input types
Text
Text (streaming)
Code
Documents (PDFs or plain-text)
Images
Video
Audio
Audio (streaming)
Output types
Text
Text (streaming)
Code
Structured output (like JSON)
Images
Audio
Audio (streaming)

Supported capabilities and features of previous generation models

These are the capabilities and featureswhen using each model withFirebase AI Logic:

	Gemini 2.0 Flash	Gemini 2.0 Flash- Lite
Thinking
Generate text from text-only or multimodal inputs
Generate images (Gemini orImagen)
Edit images (Gemini orImagen)
Generate audio
Generate structured output (like JSON)
Analyze documents (PDFs or plain-text)
Analyze images (vision)
Analyze video (vision)
Analyze audio
Multi-turn chat
Bidirectional multimodal streaming
Function calling
Code execution
Grounding with Google Search
System instructions
Count tokens

Specifications and limitations of previous generation models

These are the specifications and limitationswhen using each model withFirebase AI Logic:

Property	Gemini 2.0 Flash	Gemini 2.0 Flash- Lite
Context window* Total token limit (combined input+output)	1,048,576 tokens	1,048,576 tokens
Output token limit*	8,192 tokens	8,192 tokens
Knowledge cutoff date	June 2024	June 2024
PDFs (per request)
Max number of input PDF files**	3,000 files	3,000 files
Max number of pages per input PDF file**	1,000 pages	1,000 pages
Max size per input PDF file	50 MB	50 MB
Images (per request)
Max number ofinput images	3,000 images	3,000 images
Max number ofoutput images	---	---
Max size per input base64-encoded image	7 MB	7 MB
Video (per request)
Max number of input video files	10 files	10 files
Max length of all input video (frames only)	~60 minutes	~60 minutes
Max length of all input video (frames+audio)	~45 minutes	~45 minutes
Audio (per request)
Max number ofinput audio files	1 file	1 file
Max number ofoutput audio files	---	---
Max length of allinput audio	~8.4 hours	~8.4 hours
Max length of alloutput audio	---	---

^{*For allGemini models, a token is equivalent to about 4 characters,so 100 tokens are about 60-80 English words. ForGemini models, you candetermine the total count of tokens in your requests usingcountTokens.}

^{**PDFs are treated as images, so a single page of a PDF is treated as one image. The number of pages allowed in a request is limited to the number of images the model can support.}

Available model names of previous generation models (including retirement dates)

Model names are the explicit values that you includein your code duringinitialization of the model.

Gemini models

Gemini 2.0 Flash andGemini 2.0 Flash‑Lite models will be retired on March 31, 2026. To avoid service disruption, update to a newer model likegemini-2.5-flash-lite.Learn more.

Gemini 2.0 Flash model names

Model name	Description	Release stage	Release date	Retirement date
gemini-2.0-flash-001	Latest stable version ofGemini 2.0 Flash	Stable	2025-02-05	2026-03-31
gemini-2.0-flash	Auto-updated alias pointing to thelatest stable version ofGemini 2.0 Flash (currentlygemini-2.0-flash-001)	Stable	2025-02-10	2026-03-31

Gemini 2.0 Flash‑Lite model names

Model name	Description	Release stage	Release date	Retirement date
gemini-2.0-flash-lite-001	Latest stable version ofGemini 2.0 Flash‑Lite	Stable	2025-02-25	2026-03-31
gemini-2.0-flash-lite	Auto-updated alias pointing to thelatest stable version ofGemini 2.0 Flash‑Lite (currentlygemini-2.0-flash-lite-001)	Stable	2025-02-25	2026-03-31

Imagen models

Imagen 3 model names

Model name	Description	Release stage	Release date	Retirement date
imagen-3.0-generate-002	Latest stable version ofImagen 3	Stable	2025-01-23	No earlier than 2026-01-23
imagen-3.0-generate-0013	Initial stable version ofImagen 3	Stable	2024-07-31	No earlier than 2025-07-31

Imagen 3 Fast model names

Model name	Description	Release stage	Release date	Retirement date
imagen-3.0-fast-generate-0013	Initial stable version ofImagen 3 Fast	Stable	2024-07-31	No earlier than 2025-07-31

^{3Not supported by theGemini Developer API no matter how you access it.}

---

Next steps

Try out the capabilities of theGemini API

• Buildmulti-turn conversations (chat).
• Generate text fromtext-only prompts.
• Generate text by prompting with various file types, likeimages,PDFs,video, andaudio.
• Generatestructured output (like JSON) from both text and multimodal prompts.
• Generate images from text prompts (Gemini orImagen).
• Stream input and output (including audio) using theGemini Live API.
• Use tools (likefunction calling andgrounding with Google Search) to connect aGemini model to other parts of your app and external systems and information.