Runware
Runware offers REST and WebSocket access across image, video, audio, 3D and text. Requests use task arrays and model identifiers; pricing depends on the chosen model's rate card. Its tooling includes SDKs, a CLI, MCP and agent skills.
Platform comparison
SpicyAPI vs Runware
The closest match to us on content rules, and the one place where a competitor's public terms are about as permissive as ours. The differences are elsewhere: what a unit of billing is, and whether anything caps a runaway loop.
Starting point
What each platform is built around
SpicyAPI
SpicyAPI prices tasks in US dollars without subscription tiers. The model page estimates the selected inputs; settlement follows actual usage and that model's billing rules.
Row by row
The eight rows worth reading before you pick
Our column is the same on every one of these pages, because our behaviour does not change with who we are being compared against. Only the right-hand column is specific to this platform.
| What you are comparing | SpicyAPI | Runware |
|---|---|---|
| Content policy | No per-request content review, no safety-checker parameter to negotiate and no content-mode flag on the key. What a model will or will not produce is decided by that model. Unlawful use stays prohibited by the acceptable use policy and the reporting channel is public. | The terms prohibit harmful material involving minors and age-inappropriate material targeting minors, and we found no clause restricting other uncensored content. A separate section makes the customer responsible for age verification and access controls, and states the platform provides no consumer-facing age verification. Safety checking is an opt-in boolean and also a separately callable model; the response carries a content flag rather than refusing. |
| How you are billed | US dollars, per the unit the task itself is measured in. No credit currency to convert, no subscription, no plan tier, no monthly minimum. A quote call prices the exact body you are about to send and reserves nothing. Balance is topped up by card or crypto (USDT). | Usage is priced in US dollars. Some models have published per-output or per-second rates; others use compute time. Size, steps, numberResults and additional references can affect the charge. includeCost returns the completed task's cost. |
| A failed request | A successful task is settled at what the run used and is not refunded because the output disappointed. Failed or timed-out tasks are usually not charged. Some models charge for content refusals as stated on their model page; check those billing terms before submitting. | Documented in two places as not charged: you pay only for successful generations. An includeCost parameter returns the exact amount deducted for a given task. |
| Request shape | One create call answers 202 with a task ID; results arrive by polling or by signed webhook. Text speaks an OpenAI-compatible surface on the same key. Input is validated against a published schema with additionalProperties closed, and every write accepts an Idempotency-Key for 24 hours. | One endpoint for every modality over REST or WebSocket, with a deliveryMethod parameter choosing between a synchronous response and an async task you poll. Webhooks are supported, firing per completed item in a batch. An OpenAI-compatible chat completions endpoint is offered for text. |
| Spend controls | Set daily, monthly and lifetime spending caps on each key. New keys have no spending cap until you set one. You can also restrict a key to selected models and source IP addresses; unsettled holds count toward its allowance. | No spend cap, daily budget or per-key limit appears in the public documentation. What is documented is a prepaid balance, a reload threshold and low-balance alerts, plus read-only usage reporting that can be filtered by key. |
| Official tooling | Native REST plus an OpenAI-compatible surface, our own SDK, CLI, MCP server and Agent Skill, all on the same key. | First-party TypeScript, Python and Go SDKs, a CLI distributed as a single binary, a hosted MCP server, packaged agent skills, a Vercel AI SDK provider and a ComfyUI integration. |
| Published commitments | A published operational target of 99.9% for our own layer — accepting requests, admitting tasks, delivering callbacks. It is a target, not a service level agreement: no service credits, no compensation, and it excludes the third-party model providers that execute generation, announced maintenance windows and events outside our control. Live status is published at status.spicyapi.ai. | No uptime commitment or service level agreement appears in the public documentation. The platform states it does not enforce hard rate limits, using queue-based balancing instead. |
| Languages | Site, model documentation, console and API error text in ten languages: English, Japanese, Korean, German, French, Spanish, Brazilian Portuguese, Russian, Italian and Polish. Error codes never change with the language, so your branching stays stable. | Documentation observed in English only, with no alternate-language versions advertised. |
Swipe horizontally to see all columns.
Content policy
What the platform will refuse, and who decides
Match the AIR model identifier and version for a workflow that runs without a content filter. Optional safety checking is a separate control; it does not establish character consistency or output quality.
How you are billed
What one unit of billing actually is
Read the chosen model's rate card, including size, steps, numberResults and reference-image charges. Compare its includeCost result with your SpicyAPI quote using equivalent inputs.
Request shape
What the HTTP response means
Translate Runware's task array, taskUUID and positivePrompt into the destination endpoint's fields. Keep task correlation in your app; Runware IDs and SpicyAPI task IDs are not interchangeable.
Spend controls
What stops a loop that has gone wrong
Keep low-balance alerts for monitoring, then assign the migrated workflow a key budget. Alerts notify you; a daily cap can stop new admissions when the allowance runs out.
Fit
Which one fits the work you are doing
Pick Runware when
- You want one request shape across image, video, audio, 3D and text, and a WebSocket transport as a first-class option rather than an afterthought.
- Raw GPU access alongside the inference API matters to you — they sell it directly and we do not.
- The chosen Runware model's rate card fits your target cost at the size, duration and output count you actually need.
Pick SpicyAPI when
- You need a spend cap that is actually enforced. Their documentation describes balance alerts and a reload threshold, not a cap.
- You need USD task pricing without a subscription tier, with an estimate for the settings you choose before submitting.
- You need documentation and API error text in a language other than English.
Models
Models you can call for this
Each link opens that model family's own page: inputs, limits, content policy and the current price per unit.
ByteDanceSeedream 5.0 ProText to imageContent policy: UncensoredAlibabaQwen Image 2.1Image editingContent policy: UncensoredAlibabaWan 3.0Text to videoContent policy: UncensoredByteDanceSeedream 5.0 FlashText to imageContent policy: UncensoredAlibabaQwen Image Edit SpicyImage editingContent policy: UncensoredByteDanceSeedance 2.5Text to videoContent policy: Uncensored
FAQ
Questions about Runware and SpicyAPI
Its published terms prohibit harmful material involving minors and age-inappropriate material targeting minors. We found no clause restricting other uncensored content, and a separate section makes the customer responsible for age verification while stating the platform provides no consumer-facing verification of its own. That is close to our position. We are not going to claim their policy is stricter than it reads.
Check the chosen model, not only who hosts it. Runware documents both fixed rates and compute-time billing. Compare equivalent size, steps and output count; its includeCost reports a completed task's charge, while SpicyAPI shows an estimate before submission.
Runware's pricing documentation states in two places that failed requests are not charged. SpicyAPI: Failed or timed-out tasks are usually not charged. Some models charge for content refusals as stated on their model page; check those billing terms before submitting.
For a media integration, compare task-array handling, model availability and budget scope. Runware documents balance alerts; SpicyAPI provides per-key daily, calendar-month and lifetime caps, plus documentation and API errors in ten languages.
What we read on Runware
These are Runware's official sources. This comparison describes published features, not a benchmark or a test of their service.
Other comparisons
Next
See what a fixed per-task price looks like.
Pick a model, set the parameters you would actually send, and read the cost before you submit anything.