9 paid Sume MCP tools and what each charges, from RMBG to music
Sume's hosted MCP lists nine paid tools that need an idempotency_key. Flat prices run from $0.0225 for a cutout to $0.20 for an image upscale; model tools vary.

Sume's hosted MCP server lists nine paid tools: generate_image, generate_video, music_create, tts_create, stt_create, image_upscale_create, rmbg_create, video_upscale_create and kling-motion-control_create. Each call that creates paid work carries an idempotency_key, and if you omit payload.model the call routes to sume/auto unless the user named a family. Seven of the nine have a fixed or per-unit price on the Sume price list; the two generation tools depend on the model you choose.
That matters for an agent budget. A flat-priced tool can be costed before the call, and a model tool needs a model id and options first.
The nine tools and their prices
Prices are from the lead-verified Sume price list (read 2026-10-08) and the hosted MCP tools-and-gates docs (read 2026-10-09).
| Tool | What it creates | Price |
|---|---|---|
| generate_image | Image from the Image API catalog | Per model, for example Nano Banana 2.1 1K $0.10 |
| generate_video | Video from the video catalog | Per model and second |
| music_create | Music 1.0 track | $0.125 flat per generation |
| tts_create | Text to speech | $0.0475 per 1,000 characters |
| stt_create | Speech to text | $0.01 per audio minute |
| image_upscale_create | Larger version of an image | $0.20 per image |
| rmbg_create | Background removal | $0.0225 per image |
| video_upscale_create | Larger version of a video | $0.009 per video second, 30 s input max |
| kling-motion-control_create | Kling 3.0 Motion Control | $0.1575 per output second, 1-30 s |
Price an agent session
Take a session where an agent creates 3 images with Nano Banana 2.1 at 1K, cuts out 2 of them, upscales 1 and narrates a 600-character caption. The cost is 3 x $0.10 = $0.30 for the images, 2 x $0.0225 = $0.045 for cutouts, 1 x $0.20 = $0.20 for the upscale, and 0.6 x $0.0475 = $0.0285 for the narration. The total is $0.5735 before rounding up to the cent on each billed line.
The same session on the REST API costs the same, because the tools and the endpoints use the same price list. MCP adds the idempotency_key requirement and the OAuth consent flow, not a surcharge.
What the idempotency key protects
An agent can lose a response and retry. With a stable idempotency_key, a retry of the same operation should map to the same job, and Sume documents that a key reused for a different operation or payload returns a 409 idempotency_conflict. Build the key from the work: a SKU plus the tool name works for catalogs, and a thread id plus step number works for conversations.
Reads such as tts_source_get are free and need no key. The docs also list tools that are not on hosted MCP, including images_create and videos_create, which stay REST-only, so an agent that wants Image 1.0 behavior must call generate_image or use the Developer API.
- Name the model for generate_image if you need a predictable price
- Use one key per operation, never one key per session
- Check the 402 insufficient_credits case before long batches
Sources
Related posts
More in Developers
- Node: stream a Sume video to disk with pipeline and a short-file check
Download a finished Sume video from /v1/videos/{id}/content in Node without loading it into memory, and fail if the file is shorter than content-length.
- Omni Flash 1.1: a 3-second probe before the $3.75 4K clip
Probe a Gemini Omni Flash 1.1 prompt for $0.1125 at 360p, then spend $3.75 on the 10-second 4K clip. Price table and the max_spend_usd for each step.
- Omni reference tokens follow list order: swap list, swap meaning
<IMAGE_REF_0> is the first URL in reference_image_urls, so reordering the list rewires the prompt. A check function and an 8 s body, $1.00 at 720p on Sume.
- Omni reference videos: read Sume capabilities, not Google's note
Google's Omni page says multi-video reference is unsupported; Sume's docs list up to three reference clips. Read the catalog entry before you send them.
Written by Sume