> ## Documentation Index
> Fetch the complete documentation index at: https://docs.goframe.dev/llms.txt
> Use this file to discover all available pages before exploring further.
# AI models
> Every model in the Models picker by task, what each is best for, its options, indicative credit costs, and how to switch models per task
GoFrame Editor uses a different model for each kind of task: one to run the agent, others to generate images, video, voices, music and more. The **Models** picker lets you change any of them per project. This page lists every model with what it is good at and, where known, what it costs in credits.
## Open the picker
Click the ✦ button in the composer (tooltip **Select AI models**), or type `/model`. The **Models** dialog groups models by task (**Editing**, **Image generation** and so on); pick one per group and close. Click **Reset defaults** to return every group to its default.
Badges next to a model:
* **default**: the model used when you have not chosen another.
* **pro**: available on paid plans. On the Free plan, choosing one shows "available on paid plans" and asks you to switch to the base model or upgrade.
* **new**: recently added.
Each entry also shows a short description and either a tier word such as **Unlimited**, **Premium** or **Fast**, or a price. Credits are the currency (1 credit = \$0.01); prices and plan limits are shown in the app and may change. Generation costs are announced in the chat before a paid step runs, and when you ask for a model that is not your default the agent may check first with choice buttons (for example **Go ahead with Nano Banana 2** or **Use Gemini 3 Pro instead (my default)**); see [Review proposals and approve actions](/editor/proposals-and-permissions).
## Editing
The editing model is the agent itself: it reads your project, plans and performs the edits. The Free plan includes Claude Haiku 4.5 only.
| Model | Best for | Options | Notes |
| ---------------- | --------------------------------------------------------------------- | ------------------------------ | ---------------------------------------------------- |
| Claude Sonnet 5 | Everyday editing and agentic work; the best balance of speed and cost | Agent, Reasoning and Ask modes | **default** |
| Claude Opus 5 | Complex multi-step edits, deep reasoning, long autonomous workflows | Same modes | **pro**; premium cost per turn |
| Claude Fable 5 | The hardest creative direction and long-horizon projects | Same modes | **pro**; turns can take longer; priced above Opus |
| Claude Haiku 4.5 | Simple edits and high-volume work | Same modes | Fastest and cheapest; the only editing model on Free |
## Image generation
| Model | Best for | Options | Notes |
| ---------------- | ------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------- | -------------------------------------------- |
| Gemini 3 Pro | Best overall quality, up to 4K | Aspect 1:1, 16:9, 9:16, 4:3, 3:4, 4:5, 21:9; size 1K / 2K / 4K; up to 14 reference images; negative prompt | **default**, **pro**; \~16 credits per image |
| Nano Banana 2 | Fast, creative, iterative editing | Reference images; negative prompt | \~4 credits per image |
| Gemini 3.1 Flash | Fast generation with strong instruction following | Reference images; negative prompt | \~10 credits per image |
| FLUX.2 Pro | Photorealistic quality, precise prompt adherence, multi-reference editing | Reference images; negative prompt | **pro**; \~16 credits per image |
| FLUX.2 Turbo | Fastest FLUX; quick drafts | Text prompt only (no reference images) | \~10 credits per image |
| Seedream v5 Lite | Text rendering and complex compositions | Text prompt only | **new**; \~6 credits per image |
| Seedream v4 | Aesthetic quality and character consistency | Text prompt only | \~6 credits per image |
See [Generate images](/editor/generate-images).
## Video generation
Costs are per second of generated video.
| Model | Best for | Options | Notes |
| ----------------- | ------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------- |
| Veo 3 Fast | The most capable all-round model; native audio | 4 / 6 / 8 s; 16:9 or 9:16; 720p or 1080p; image-to-video, first + last frame; up to 3 reference images ("Ingredients") | **default**, **pro**; \~23 credits/s |
| Kling 3.0 Pro | Cinematic quality, motion physics, precise camera control | 5 / 10 s; 16:9, 9:16, 1:1; optional audio; Custom Elements and Multi-Shot for multi-segment narratives | **pro**; \~75 credits/s |
| Kling O3 Pro | The most versatile Kling: reference-to-video, video editing and restyling | 5 / 10 s; 16:9, 9:16, 1:1; audio; video-to-video | **pro**, **new**; \~75 credits/s |
| Kling 2.5 Turbo | Fast, affordable clips | 5 / 10 s; 16:9, 9:16, 1:1; silent (no audio) | **pro**; \~23 credits/s |
| Seedance 1.0 Lite | Budget clips with native audio | 5 / 10 s; 16:9, 9:16, 1:1; 1 reference image; audio | **pro**; \~23 credits/s |
| Seedance 2.0 | Multi-segment narratives: story, adaptation, montage | 4–15 s or auto; 1:1, 3:4, 4:3, 9:16, 16:9, 21:9; 480p or 720p; audio; up to 9 image + 3 video + 3 audio references | **pro**, **new**; \~38 credits/s |
| Seedance 2.0 Fast | Preview iterations before a final render with Seedance 2.0 | Same options as Seedance 2.0 | **pro**, **new**; \~30 credits/s, lower latency |
See [Generate video](/editor/generate-video).
## Avatar video
| Model | Best for | Options | Notes |
| ---------------------------- | ---------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------- |
| HeyGen Avatar 4 Digital Twin | Premium presenters from 600+ pre-made avatars, natural lip-sync and gestures | Avatar style normal / circle / close-up; 70+ voices and 100+ languages; speed; talking style stable / expressive; resolution 360p–1080p; background color, image or video; optional captions | **default**, **pro**; billed per second |
| HeyGen Photo Avatar 4 | Turning any portrait photo into a speaking avatar; personalized messages | Same options, from your own photo | **pro**, **new**; billed per second |
See [Avatar videos](/editor/avatar-video).
## Image upscaling
| Model | Best for | Options | Notes |
| ------------- | ----------------------------------------- | ------------------ | --------------------------------------------------------------- |
| SeedVR2 | High-quality upscaling up to 10K px | Scale factor 1–10× | **default**, **new** |
| Bria Creative | Stylized images; adds AI-generated detail | Per-image | Better when the source is illustrative rather than photographic |
## Video upscaling
| Model | Best for | Options | Notes |
| -------------- | ----------------------------------------------------------------- | ----------------------------------------------- | -------------------------- |
| SeedVR2 | Fast upscaling up to 4K with the same engine as the image version | Targets up to 4K | **default**, **new** |
| Topaz Video AI | Highest fidelity for final production output | Targets 720p to 4K; frame rate 24 / 30 / 60 fps | Billed per second of video |
See [Upscale images and video](/editor/upscale).
## Text to speech
Indicatively about 4 credits per generation.
| Model | Best for | Options | Notes |
| --------------------- | ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------- |
| ElevenLabs Eleven v3 | Expressive, natural narration; supports voice cloning | Voices Rachel (default), Domi, Bella, Elli, Charlotte, Grace, Dorothy, Josh, Arnold, Adam, Sam, Michael, Clyde, or a cloned voice; speed 0.5–2.0; stability | **default** |
| ElevenLabs Turbo v2.5 | Longer scripts and drafts, faster and cheaper | Same voices and controls | |
| Minimax Speech HD | Naturalness and emotional range | Voices Wise\_Woman (default), Friendly\_Person, Inspirational\_Girl, Deep\_Voice\_Man, Calm\_Man, Casual\_Guy, Lively\_Girl; speed | |
| Minimax Speech Turbo | Longer scripts, faster and cheaper | Same voices | |
Voice cloning needs a sample of at least 10 seconds and the speaker's consent. See [Voiceover and text-to-speech](/editor/voiceover).
## Sound effects
| Model | Best for | Options | Notes |
| -------------- | ---------------------------------------- | ------------------------------- | -------------------------------------------- |
| ElevenLabs SFX | Any sound effect from a text description | Duration 0.5–22 s (default 5 s) | **default**; runs through the `/voice` skill |
## Video to SFX
| Model | Best for | Options | Notes |
| ------------- | ------------------------------------ | --------------------------------------- | ----------- |
| MMAudio | Synchronized sound for a silent clip | Optional text prompt to steer the sound | **default** |
| Mirelo AI SFX | Cinematic sound design | Video input only | **pro** |
See [Music and sound effects](/editor/music-and-sfx).
## Text to music
About 20 credits per generation.
| Model | Best for | Options | Notes |
| ------------- | ------------------------------------------------------- | ----------------------------- | ----------- |
| Minimax Music | Background music in many styles; songs with your lyrics | Style prompt; optional lyrics | **default** |
| CassetteAI | Fine-grained style control; loop-ready beds | Up to 120 s | |
| Lyria 2 | Cinematic composition and musical quality | About 30 s per generation | |
## Generation settings
**Aspect ratio** (**Auto**, 16:9, 9:16, 1:1, 4:3, 3:4, 4:5, 21:9), **Duration** and **Audio** (**Auto**, **On**, **Off**) for image and video prompts are set from the **Generation settings** button (sliders) in the composer. The duration choices follow the selected video model (4 / 6 / 8 s for Veo 3 Fast, 5 / 10 s for most others, 4–15 s for Seedance 2.0). See [Write prompts, references and commands](/editor/prompting).
## Continue
* [Generate images](/editor/generate-images)
* [Generate video](/editor/generate-video)
* [Voiceover and text-to-speech](/editor/voiceover)
* [Plans and credits](/editor/plans-and-credits)