MundoKit

AI Video Input Token Calculator

Enter a video duration and sampling plan, or a measured agentic scenario, to check a token budget.

Inputs
Seconds; fractional seconds accepted.
Planning assumption; configure matching FPS in your API request.
Blank uses the selected resolution preset, checked 2026-11-14.
Used only in static mode with audio enabled; checked 2026-11-14.
User estimate. Zero excludes metadata; total is then incomplete.
Example budget, not a model limit.
Example only; enter your model’s actual context budget.

Results

Estimated input tokens
5,880
Input plus output reserve
6,880
Context remaining
993,120
Token components
ComponentTokens
Sampled frames60
Visual tokens3,960
Audio tokens1,920
Prompt/metadata allowance0
  • Within the entered context budget.
  • Static estimates exclude unknown metadata unless you add it. Agentic scenarios do not predict navigation or billable thought tokens.

Results update as you type. The share link stores your inputs after the # in the address, which is not sent to our server, but anyone you send it to can read them.

How it works

Google’s static video documentation gives 66 tokens per low-resolution frame (the API default), 258 otherwise, and 32 audio tokens per second. These presets were checked on 2026-11-14 and can be overridden. Choose the resolution used in your request.

input = ceil(seconds × FPS) × tokens/frame + seconds × audio tokens/second + allowance

Frame counts round up as a planning convention; real sampling boundaries and metadata may differ. Add prompt/metadata allowance and output reserve before comparing with your entered context budget. Agentic mode uses your entered loaded input tokens; navigation and thought usage need separate measurement. No intermediate token totals are rounded; display uses up to two decimals.

Worked examples

One minute with audio at the API default

60 seconds at 1 FPS and 66 tokens/frame uses 3,960 visual tokens plus 1,920 audio tokens = 5,880 input tokens before metadata. A 1,000-token reserve makes 6,880.

Low resolution without audio

60 seconds at 1 FPS and 66 tokens/frame, without audio, uses 3,960 input tokens. With a 1,000-token reserve it uses 4,960 tokens of the entered context budget.

Selected 258-token alternative

At 60 seconds, 1 FPS and 258 tokens/frame with audio, input is 17,400 tokens before metadata; a 1,000-token reserve makes 18,400.

Frequently asked questions

Is agentic video a fixed tokens-per-second rate?
No. Enter measured loaded input tokens; content and navigation change usage.
Does resolution mean pixel dimensions?
Here it means the API media-resolution preset; this tool does not calculate image tiling from pixels.
Is metadata included?
Only when you enter an allowance. Zero allowance omits it.
Is this a price calculator?
No. Input context tokens and billed thought/output tokens can have different prices.

Sources

Last reviewed 2026-11-14; next review due by 2027-02-12. The method is written out above so you can check our math.