AI Models
By default, the platform selects an AI model for each request based on what the request needs and the user's plan. Free accounts use lite models. Basic can use lite and standard models. Pro and Ultra can use every model tier — more capable models cost more credits. Most apps do not need to configure models manually.
Default Selection
When an app makes an AI request without specifying a model, the platform chooses an appropriate available model. The selected model depends on the request type, required capabilities, your account preference if you have one, and efficient defaults for your plan.
Free accounts are limited to lite models. Basic can use lite and standard models. Pro and Ultra can use lite, standard, and advanced models, including explicit selectors in apps that expose them. Plans also differ in monthly credit allowance, concurrency, and default model selection.
Account Preferences
You can set default models for text, images, video, and speech in account settings. These defaults apply across the apps you use when an app does not request a specific model.
Choose Automatic to follow the platform default for your plan. Choose a specific model to try that model first.
Account preferences are not a hard pin. If your preferred model cannot handle a request — for example, it does not support an attachment type or web search — the platform can fall back to another available model, the same way Automatic does. An app that pins a model still takes priority over your account default.
Model Tiers
Models are grouped into three tiers:
| Tier | Best for | Credit cost |
|---|---|---|
| Lite | Fast, lower-cost tasks such as formatting, classification, and quick summaries | Lowest per request |
| Standard | Balanced quality and cost for most everyday tasks | Moderate per request |
| Advanced | Complex reasoning, higher-quality writing, and demanding tasks | Highest per request |
Fallback Behavior
Automatic model selection includes graceful fallback. If the first model cannot handle a requested capability, such as a specific attachment type or web search, the platform can try another available model so the request still succeeds when possible.
Fallback applies when the platform is choosing the model: Automatic, or your account default. If an app pins a specific model, the platform uses that model and returns an error if it cannot satisfy the request.
Model Types
The platform supports models for text, images, video, and speech. For developer-facing details, see the AI text, AI image, AI video, and AI speech references.
To customize model behavior in an app, see Model Selection.
Available Models
Here are all the models currently available on the platform, grouped by modality. Use the Model ID value when exposing a model selector or pinning a specific model.
Text Models
Used by useAIChat, useAIText, and useAIObject for conversations, text generation, and structured data extraction.
| Model | Model ID | Provider | Tier | Description |
|---|---|---|---|---|
| GPT-6 Luna | gpt-6-luna | OpenAI | Lite | OpenAI's most efficient model for focused, high-volume tasks. |
| Gemini 3.5 Flash Lite | gemini-3.5-flash-lite | Lite | Google's most efficient model. Ultra-fast and cost-effective. | |
| Claude Haiku 4.5 | claude-haiku-4-5 | Anthropic | Lite | Anthropic's fastest model with near-frontier intelligence. |
| Gemini 3.8 Flash | gemini-3.8-flash | Standard | Google's most intelligent Flash model. Fast, capable, and built for agents and long-horizon work. | |
| Grok 4.6 | x-ai/grok-4.6 | xAI | Standard | xAI's smartest model. Frontier coding, knowledge work, and STEM. |
| GPT-5.6 Terra | gpt-5.6-terra | OpenAI | Standard | OpenAI's GPT-5.6 model that balances intelligence and cost. |
| Claude Sonnet 5 | claude-sonnet-5 | Anthropic | Standard | Anthropic's mid tier. Strong accuracy, fast. |
| Gemini 3.1 Pro | gemini-3.1-pro-preview | Advanced | Google's best model with world-class multimodal understanding. | |
| GPT-6 Sol | gpt-6-sol | OpenAI | Advanced | OpenAI's model built for complex coding and agentic workflows. |
| GPT-6 Astra | gpt-6-astra | OpenAI | Advanced | OpenAI's most capable model, built for the hardest end-to-end work. |
| Claude Opus 5.5 | claude-opus-5-5 | Anthropic | Advanced | Anthropic's model for long-running agentic coding and knowledge work. |
| Claude Fable 5.1 | claude-fable-5-1 | Anthropic | Advanced | Anthropic's most capable model. Best for demanding, long-horizon work. |
Image Models
Used for generating and editing images within your apps.
| Model | Model ID | Provider | Tier | Description |
|---|---|---|---|---|
| GPT Image 2.5 Flare | openai/gpt-image-2.5/flare | OpenAI | Lite | OpenAI's latest image model. Extremely detailed with fine typography. |
| Grok Imagine Image 2.0 | xai/grok-imagine-image/v2.0 | xAI | Lite | xAI's image model for generation and editing. |
| Gemini 3.1 Flash Image (Nano Banana 2) | fal-ai/nano-banana-2 | Standard | Google's fast tier. Quick, good quality. | |
| Gemini 3 Pro Image (Nano Banana Pro) | fal-ai/nano-banana-pro | Advanced | Google's flagship. Best quality, higher res. |
Video Models
Used for generating videos from text or images.
| Model | Model ID | Provider | Tier | Description |
|---|---|---|---|---|
| Gemini Omni Flash 1.1 | google/gemini-omni-flash/v1.1 | Lite | Google's multimodal video model with native audio, image animation, reference images, and editing. | |
| Grok Imagine Video 1.5 | xai/grok-imagine-video/v1.5 | xAI | Lite | xAI's latest video model with native audio, 1080p generation, image animation, and reference guidance. |
| Seedance 2.0 Fast | bytedance/seedance-2.0/fast | ByteDance | Standard | ByteDance's faster Seedance 2.0 video model with native audio, image animation, and reference guidance. |
| Kling V3 | fal-ai/kling-video/v3/standard | Kling | Standard | Fast cinematic video generation with native multi-shot support. Powered by Kling 3.0. |
| Kling O3 | fal-ai/kling-video/o3/standard | Kling | Standard | Kling's faster O3 model for text-to-video, image animation, video references, and video editing. |
| Seedance 2.0 | bytedance/seedance-2.0 | ByteDance | Advanced | ByteDance's cinematic video model with native audio, image animation, and reference-image guidance. |
| Seedance 2.5 | bytedance/seedance-2.5 | ByteDance | Advanced | ByteDance's latest cinematic video model with native audio, 30-second generation, and multimodal references. |
| Kling V3 Pro | fal-ai/kling-video/v3/pro | Kling | Advanced | Cinematic video generation with fluid motion, native audio, and multi-shot support. Powered by Kling 3.0. |
| Kling O3 Pro | fal-ai/kling-video/o3/pro | Kling | Advanced | Kling's O3 Pro model for text-to-video, image animation, video references, and video editing. |
Gemini Omni Flash 1.1 resolutions
Use model ID google/gemini-omni-flash/v1.1. The previous google/gemini-omni-flash ID automatically resolves to version 1.1.
All supported workflows—text-to-video, image animation, reference-image generation, and video editing—accept these resolutions, with native audio included:
| Resolution value | fal cost per output second | AI credits per output second |
|---|---|---|
360p | $0.03 | 30 |
720p | $0.10 | 100 |
1080p | $0.15 | 150 |
4k | $0.30 | 300 |
The model default is 720p. Quality settings, including your plan's default quality, select 360p for low, 720p for medium, and 4K for high. Pass an explicit resolution to override quality, such as resolution: "1080p". An eight-second 1080p video uses 1,200 AI credits.
Prices verified against fal's model pricing on September 22, 2026.
Speech Models
Used for text-to-speech generation.
| Model | Model ID | Provider | Tier | Description |
|---|---|---|---|---|
| GPT-4o Mini TTS | gpt-4o-mini-tts | OpenAI | Standard | OpenAI's fast tier. Natural, cost-effective. |
| ElevenLabs Turbo v2.5 | fal-ai/elevenlabs/tts/turbo-v2.5 | ElevenLabs | Standard | ElevenLabs' low-latency model. Fast, natural speech in 32 languages. |
| TTS-1 HD | tts-1-hd | OpenAI | Advanced | OpenAI's highest-quality TTS model. |
| Eleven v3 | fal-ai/elevenlabs/tts/eleven-v3 | ElevenLabs | Advanced | ElevenLabs' most expressive model. Rich emotion, 70+ languages. |