ChatCut credits pay for AI processing and media generation. Manual timeline editing, uploads, project browsing, transcription, and local export do not consume credits. Cloud export may use credits after the project owner’s free allowance. A manual export in the web editor shows the estimated credits and asks you to confirm any charge before it starts; a cloud video or audio export that an agent submits starts without a separate confirmation (see Web vs Desktop Export). A plan can still limit access to a feature even when you have enough credits.
Can I estimate the cost from 30–60 minutes of raw footage?
Raw footage length alone cannot tell you whether 100 credits will cover an edit. Uploading and manually editing that footage do not consume credits. Asking the built-in ChatCut Agent to edit it uses variable model usage; generating new media adds the charges for those generation actions. Revisions and the context processed can also change the total.
Before starting, separate the work you need: editing existing footage, asking the Agent for changes, and generating new clips, images, or audio. Check your current plan and balance in Settings › Billings and plan. For video generation, review the estimate beside Video Gen after choosing its settings. Some Agent-driven generation requests show a confirmation card; do not assume every action has a fixed price preview.
After an action completes, check Credit activity → Detail for its actual deduction. A cost shown in someone else’s demonstration is not a quote for your project. If you use a Codex or Claude plugin, also account for the host agent’s separate allowance, as explained below.
Which actions use ChatCut credits?
| Action | How it is charged |
|---|---|
| ChatCut Agent messages in the web or desktop editor | Variable. The model, input and output tokens, cached context, conversation length, and project context all affect the charge. |
| AI video generation | By generated seconds, model, and resolution. Seedance models also bill by frame size and any reference or source video. See the table below. |
| AI image generation | By model, output quality, resolution, and image count. |
| Motion graphics created by the built-in Agent | Variable, based on the AI model usage required to create or revise the graphic. In ChatCut Desktop, the Agent writes the graphic itself, so the cost is part of its message usage. |
| AI avatar videos | By generated seconds, avatar type, and resolution. Creating a custom avatar also uses credits. |
| Video translation and lip sync | By video length; video translation also depends on the translation mode you choose. |
| AI music | About 0.2 credits per instrumental track; a song with vocals uses about 0.6–0.72 credits. |
| AI sound effects | 0.144 credits per generated second. |
| AI voiceover | 0.80 credits per 1,000 characters; ElevenLabs voices use 0.96 credits per 1,000 characters. |
The estimate beside the Video Gen send button updates after you choose the model, resolution, duration, and supported video inputs. Agent-driven video, image, or motion-graphics generation may instead show a confirmation card before the job starts.
How can I keep Agent conversation usage focused?
Agent-message charges vary with the context ChatCut must process. To avoid paying for unnecessary back-and-forth:
- Put the goal, required source material, output format, and constraints in one specific request.
- Use attachments or Selection Mode references instead of repeatedly describing which asset, clip, time, screen region, or transcript passage you mean.
- Keep related revisions in the same conversation while the previous context is useful.
- Start a new conversation for an unrelated task when the old discussion is no longer needed. Create a separate project only when the media and Timeline should also be separate.
These steps reduce unnecessary context; they do not promise a fixed charge. Verify the completed amount in Credit activity → Detail.
How many credits does video generation use?
Rates on this page apply to accounts on ChatCut’s current pricing; if your subscription started under earlier pricing, the editor may show lower rates while that subscription stays active. A completed job is rounded up to the nearest 0.01 credit, so multiplying a displayed per-second figure can differ by a few hundredths from the final transaction.
| Model and mode | 480p | 720p | 1080p |
|---|---|---|---|
| Seedance 2.5 — output only | about 0.4811/sec | about 1.0816/sec | about 2.677/sec |
| Seedance 2.5 Flash — output only | about 0.3161/sec | about 0.7108/sec | Not available |
| Seedance 2.0 | about 0.3305/sec | about 0.7108/sec | about 1.773/sec |
| Seedance 2.0 Fast | about 0.2658/sec | about 0.5717/sec | Not available |
| Seedance 2.0 mini | about 0.1653/sec | about 0.3554/sec | Not available |
| Kling 3.0 | — | 0.6048/sec | 0.8064/sec |
| Kling 4.0 | — | 0.7536/sec | 1.128/sec |
| Kling 4.0 Flash | — | 0.4512/sec | Not available |
| Gemini Omni | — | about 0.4865 per actual generated second | about 0.7298 per actual generated second |
Seedance 2.0 and Seedance 2.5 also offer native 4K. On Seedance 2.0, 4K costs about twice the 1080p rate. On Seedance 2.5, 4K output costs about 4.311/sec, about 1.6 times its 1080p rate. The editor shows the exact estimate before you send.
Seedance models are billed by the provider’s token usage, which depends on the output frame size and length. The Seedance rows are approximate output-only rates for a 16:9 frame; other aspect ratios can cost slightly more or less. Adding a reference or source video to any Seedance model adds input usage based on its duration and resolution. The editor includes that input in the preflight estimate; final billing uses the provider-reported usage when it is available.
For Kling 3.0, attaching reference images applies Kling’s lower Omni rate: 0.5376/sec at 720p and 0.672/sec at 1080p. For Kling 4.0 with reference videos, billed seconds are the output length plus the longer of the output and the total reference-video length, charged at 0.4512/sec (720p) or 0.6768/sec (1080p). Kling 4.0 Flash with reference videos bills seconds the same way at 0.3024/sec. Kling rates are the same for subscriptions on earlier pricing.
For Gemini Omni, the requested duration is a target. Billing follows the duration actually returned by the model. Gemini Omni also offers 360p (about 0.1622/sec) and 4K (about 1.4596/sec).
MiniMax H3 (768p and 2K) and MiniMax H3 Max (480p and 768p) are billed per generated second by resolution. On MiniMax H3, additional reference images can add to the charge. The editor shows the estimate before you send.
If the estimate in the editor differs from this table, use the editor estimate before submitting and confirm the completed transaction afterward.
What does not use ChatCut credits?
- Uploading files, organizing the Media Pool, and browsing projects.
- Manual editing such as trimming, moving, styling, and arranging timeline items.
- Transcription and standard caption editing.
- Local video, audio, subtitle, and XML export, in the browser or in ChatCut Desktop. Cloud export may use credits after the project owner’s free allowance. In the web editor, ChatCut asks you to confirm any charge before a manual export starts. A cloud video or audio export that an agent submits starts without a separate confirmation; if you don’t want to spend credits, ask the agent to use browser export instead, or export locally in ChatCut Desktop. Depending on your plan, 4K export may require an upgrade; ChatCut shows an upgrade prompt if it does. See Web vs Desktop Export and Export Panel.
These are free when you do them yourself in the editor. When you ask the built-in ChatCut Agent to do any of them, its messages use credits, listed as AI Agent in Credit activity.
What changes when I use a Codex or Claude plugin?
When Codex, Claude, or another supported local agent controls ChatCut through the plugin or MCP connection, the host agent’s reasoning uses that host’s own subscription or API allowance. ChatCut credits are used only when the workflow calls a ChatCut-hosted generation service, such as video, music, sound effects, or voiceover generation.
If Codex uses its own image-generation capability, that image generation follows your ChatGPT/Codex allowance instead of creating a ChatCut image-generation charge. Check the confirmation shown by the active tool when a workflow mixes host-provided and ChatCut-provided generation.
How do generation confirmations work?
A confirmation card can show Allow this time, Auto-allow this generation category in this project, and Cancel this generation. For Motion Graphics, the auto-allow option reads Auto-allow this generation category in all projects.
To change a saved choice on the web, open Agent settings → Auto-allow generation. In ChatCut Desktop, click the agent button at the bottom of the AI composer; under Settings, open Auto-allow generation. See Generation Confirmations and Auto-Allow.
How do I verify the final charge?
- On the web, open the avatar menu in the top-right of the editor and select Settings, then Billings and plan. In ChatCut Desktop, open the account menu at the bottom-left of the sidebar and select Credits.
- Find Credit activity on the Credits tab.
- Open Detail.
- Read the newest negative transaction and the balance shown after it.
The transaction list is the source of truth for completed charges. If a generation fails, open its card in the Media Pool, then check Credit activity instead of assuming that the estimate was deducted.