Guides
Sume guides: the API, Formats, agents, models, media tools and pricing, each answer dated and sourced.
Qwen Image 2.1 Pro vs Turbo: price, steps and rate limits
Turbo is 8 steps at $0.016 per image; Pro is $0.04 with a 20 RPM cap. Both are on Alibaba Model Studio; Sume lists neither yet.
Qwen Image 2.1 rate limits: 120 RPM Turbo, 20 RPM Pro, and batching
Alibaba caps qwen-image-2.1-turbo at 120 requests per minute and Pro at 20. Here is how to size a batch and where Sume's async mode fits.
Qwen Image 2.1 Turbo vs FLUX.2 pro vs Seedream 5 Lite: price per image
Vendor list prices are $0.016 (Qwen Turbo, Singapore), $0.03 (FLUX.2 pro) and $0.035 (Seedream 5.0 Lite). Sume lists the last two.
Real-time world model or video API? A 24 fps session vs a job
HelixWorld runs interactively at 24 fps on one GPU. Sume video is an async job you poll or receive by webhook. How to choose, and what each costs.
Repeatable AI images: seed returns 400 on Sume, so do this
Sume's schema accepts seed, but no image model lists it, so seed returns 400 unsupported_parameter. Pin the model and prompt, and keep the output file.
Seedance 15-second clip price ladder on Sume: Mini to 2.5
A 15-second Seedance clip on Sume costs $1.32 on Mini at 480p up to $21.33 on 2.5 at 1080p. Four models, three tiers, list x 1.25, read 2026-10-13.
Seedance 2.5 pro tools vs the Sume API: which fields exist
Sume exposes Seedance 2.5 by prompt, frames, references, duration and resolution. No white-model, green-screen or blocking field is documented. Read 2026-10-13
Seedance 2.5 vs Gemini Omni Flash: length, size and inputs
Seedance 2.5 on Sume runs 4-30 s up to 1080p; Omni Flash 1.1 runs 3-10 s up to 4K and adds edits. Pick by length first. Read 2026-10-13.
Seedream 5.0 Lite vs 4.5 vs 4.0 on Sume: prices and ratios
Sume bills Seedream 5.0 Lite $0.04375, Seedream 4.5 $0.05 and Seedream 4.0 $0.0325 per image. All take 10 references; only 4.0 lists auto.
Sonic and Eleven v4 share 44 language codes on Sume
Sume's TTS Router lists the same 44 language codes for Sonic 3 and for Eleven v3 and v4, so the language field alone does not pick the engine.
Image model sume/auto or a pinned id: when to send which
Send sume/auto when you want Sume to choose and pin an id when you need the same look and price every time. Auto never names the model that ran.
sume/auto or a fixed video model id: decision rules for client work
Use sume/auto for 3-10 s clips with no family requirement. Pin an id when you need over 10 s, 1080p-only, a given family or 4:3. Read 2026-10-13.
Sume TTS Router: Eleven rows have no timestamps or avatar voice
On Sume's TTS Router, Eleven rows refuse timestamps, segmentation, transcript_source and avatar selectors. Sonic rows accept them. Pick by the fields you need.
Sume TTS Router output formats: WAV on Sonic, MP3 on Eleven
Sume's Sonic rows return mp3 or wav up to 48 kHz. Eleven rows return mp3 only, and three of them only mp3_44100_128. Pick by the format you need.
Sume TTS Router voices: the sonic and eleven families
GET /v1/tts-router/voices lists voices for one family, sonic or eleven. A model speaks only with voices from its own family, so pick the model first.
AI video models with 4:3 and 3:4 on Sume: which rows list them
For 4:3 or 3:4 video on Sume, use a Seedance, Wan 3.0 or H3 row. Kling 3, Omni and Veo do not list those ratios. Read 2026-10-13.
Sume webhook Send test vs Redeliver: only one replays a real job
Send test posts a dummy webhook.test payload to a URL you type. Redeliver re-sends the real terminal event for one job with a fresh signature. Do not swap them.
Sume rejects localhost webhook URLs: test locally with a poll fallback
Sume only accepts public HTTPS webhook URLs and rejects localhost and private networks. For local work, poll the job and keep webhooks for a deployed endpoint.
What list x 1.25 adds per second on Sume video, by model
Sume's 1.25 margin adds $0.0125 to $0.2843 per second on the rows in this table. Table of list, margin and total, read 2026-10-13.
TTS language en-US vs en-GB on Sume: regional tags compare by language
For the voice-language check, Sume compares regional tags by primary language, so en-GB on an en-US voice does not trigger the 409, and fil matches tl.
Live commerce TTS rejects literal text: use transcript_source
Sume's live-commerce TTS takes only transcript_source (a script revision plus sentence IDs) and rejects literal transcript text, even if correct.
TTS price per 1,000 characters: every Sume row next to MAI-Voice
Sume's TTS Router bills $0.0475 per 1,000 characters on Sonic rows and $0.05 to $0.125 on Eleven rows; Microsoft lists MAI-Voice-2.1 at $0.022.
Veo 3.1 on Sume: Google Cloud rows, 720p, text-to-video only
Veo 3.1, Fast and Lite appear on Sume only where Google Cloud serves them: text-to-video, 720p, 4, 6 or 8 s, 16:9 or 9:16. Check your own list. Read 2026-10-13.
Video editing API or AI prompt edit: which Sume endpoint to call
Trim, frames, captions and filter are deterministic jobs; Omni edit changes content from a prompt. Pick the cheapest route that does the change you need.
video_url on /v1/videos fails: edits go to /v1/video-router/generate
Sume's /v1/videos has frame_images and input_references but no edit field. Send video_url to /v1/video-router/generate with gemini-omni-flash-1.1.
WAND-Vega for e-commerce video: what Tencent says, what Sume lists
Tencent describes WAND-Vega as trained for vertical industries such as e-commerce. On Sume it is three text-to-video rows, 4-15 s, up to 4K. Read 2026-10-13.
What balance do 100 video ads need? Sume reserves at submit
Sume reserves the credits of every accepted job at submit, so a 100-ad batch needs the whole total in the balance. 402 insufficient_credits otherwise.
Which video_url row to call: Omni edit, Recast or Motion Transfer
Three Sume rows accept video_url: gemini-omni-flash-1.1 edits by prompt, h3-max-recast swaps people, higgsfield-genjutsu moves motion. Limits and prices.
Whistle's 86,174-utterance benchmark: test your own clips first
Cactus Whistle's card reports word error rates over 86,174 utterances. Score 20 of your own clips against a human transcript before choosing it over hosted STT.
Whistle needs 16 kHz mono audio: extract it from a video first
Whistle takes 16 kHz mono audio. Sume audio detach returns exactly that shape from a hosted video with sample_rate 16000 and channels mono.
Whistle speech model: Apache 2.0, 16.9 MB, offline or Sume STT?
Cactus Whistle is a 16.9 MB Apache 2.0 speech-to-text model for seven languages. Run it on-device, or send the clip to Sume STT for other languages.
Why a video model is missing from your Sume model list
A documented video id can be absent from the model list: Veo needs Google Cloud, Tencent rows need the Tencent switch. Absent ids 404. Read 2026-10-13.
Astro 7.3.8 actions: call the Sume video API, key stays on the server
Astro actions give a browser form a typed way to start a Sume video job and poll it, without the API key in client code. Zod 4 validation, 402 handling.
Copilot CLI 1.0.97 Autopilot denies a hook's ask: paid Sume calls
Copilot CLI 1.0.97-0 makes Autopilot deny a preToolUse hook's ask instead of stalling. What that means for a paid Sume MCP call, and which gates still apply.
Copilot CLI 1.0.97 rechecks cached MCP tool definitions: Sume
Copilot CLI 1.0.97-0 rechecks a cached MCP tool call when a tool-list refresh changes the catalog. How to use Sume's tools_schema so arguments stay valid.
Copilot CLI 1.0.97 resume reports interrupted calls: Sume retries
Copilot CLI 1.0.97-0 reports interrupted tool calls on resume and avoids replaying canceled writes. Reuse the same idempotency_key for an interrupted Sume call.
Copilot CLI 1.0.95 and 1.0.96 sandbox secrets: where Sume's key fits
Copilot CLI 1.0.95 and 1.0.96 add sandbox injectHosts keys and masking hosts. Sume's MCP works with OAuth or an API key header; here is how to choose.
Effect 4.0.4 retryTransient honors Retry-After: key your Sume submit
Effect 4.0.4 makes HttpClient.retryTransient honor Retry-After, so a Sume 429 waits as told. It also retries POST on 5xx, so send an Idempotency-Key.
Flood insurance agent video: the coverage gap and 30-day wait
An independent agency can explain that homeowners policies skip flooding with a 30-second avatar clip. Script from Ready.gov, Sume costs at two tiers.
Four-panel comic with your own character: an API recipe
Recraft's Oct 9 comic guide, as an API job: draft four panels at 0.5K, render finals with the hero as a reference, letter the text yourself.
Free in-browser video upscaler (10 files, 5 min) vs an upscale API
A free upscaler added batch mode on Oct 11: 10 files, 100 MB, 5 minutes each, in-browser. When Sume's video upscale API fits better, and what it costs.
Go 1.27.2 synctest: test a 30-second Sume poll loop in milliseconds
Test a Go poll loop for Sume video jobs with testing/synctest: 90 seconds of fake sleeps and a 20-minute deadline pass in under a second, no sockets.
Higgsfield AI Influencer's 19 settings vs a Sume avatar and face swap
Higgsfield builds a character from 19 settings, then drops it into video. Sume makes an avatar from a prompt, photo or 3 traits, then script or face-swap it.
k6 2.3 --once and --scenario: smoke-test the Sume API without spend
Use k6 --once and --scenario for a Sume API smoke test that costs nothing: models, a deliberate 400, and a paced balance read that counts 429s.
One hero image to 3D toy, anime and pixel sprite via API
Recraft's Oct 9 art style challenge as three Sume calls: attach the hero, fix silhouette and palette, vary only the medium. Ratios and costs.
OpenCode 2.0.28 auto-accept per project: paid Sume tools
OpenCode 2.0.28 lets you override Auto-accept permissions for a single project. Use Sume's read-only OAuth scope and spend gates before turning it on.
Place your character in concept art scenes with two references
Design locations first, then put the hero in each one with two input_references. Recraft's Oct 9 method as a Sume API script with costs.
Playwright 1.64 lock and --shuffle for a Sume API test suite
Playwright 1.64 adds a describe lock and --shuffle. Use them to test Sume video calls: free 400 checks run in parallel, one paid test holds a credits lock.