A 5,000-character voiceover costs $0.2375: sizing the run cap

Sume TTS costs $0.0475 per 1,000 characters, so a 5,000-character voiceover is $0.2375 and a 20,000-character script is $0.95. For an Agent Completion that only narrates, a generation_spend_cap_usd of 0.5 covers up to 10,000 characters.

Cost by script length

The price is linear in characters, so scale the table to your script.

Send it

Pass the script as input data and keep the instruction short. input is written to a file as data, not treated as instructions, so a script that says something odd cannot rewrite the task.

{
  "instruction": "Narrate the script in input.script as one voiceover.",
  "input": { "script": "..." },
  "generation_spend_cap_usd": 0.5
}

Limits and gotchas

The table covers narration only. If the same completion also makes images or video, add their prices to the cap.

• Keep input a JSON object; a malformed input is a 400 invalid_request.
• Poll the run at its status_url; media URLs on completion are durable media.sume.com links.
• Splitting a script across several runs costs the same, because the price is per character.