ChinaAPI
Sign up
English
English 日本語 (English only) 한국어 (English only) ไทย (English only) Tiếng Việt (English only) Русский (English only) Bahasa Indonesia (English only) Português do Brasil (English only) Español (English only) Français (English only)
Sign up

Start from the outcome

Choose a task, not a leaderboard.

Each workflow defines the input, compatible executors, cost drivers, lifecycle, recovery behavior, and acceptance check before you create an account.

Open image-to-video
7task paths1locked contract0executable workflowsRead-onlycontract pre-flight

Workflow catalog

What are you trying to finish?

Animate an imageReference image → controlled motion → videoImage · motion · asyncGenerate a videoPrompt → video capability pathsPrompt · duration · ratioCreate product imagesProduct reference → campaign variationsReference · batch · editEdit an imageImage + instruction → production assetMask · reference · outputRoute an Agent taskGoal → model constraints → recoverable callSchema · budget · fallbackTranscribe audioAudio → timed, structured transcriptAudio · language · timestampsBuild an AI Chinese speaking tutorVoice turn → transcript → coaching → spoken replyFlagship project · Agent recipe
Every workflow answers the expensive questions first.Required inputSafe defaultsCost driversUnsupported combinationsRecovery pathAcceptance check
MandarinSpark · project boundary and build path

Featured project · MandarinSpark

Build an AI Chinese speaking tutor

MandarinSpark is the owned reference project for composing ASR, an LLM coach, deterministic pinyin, and TTS. Production integration remains a separate learner-app delivery: every model must first be registered, routed, priced, and fixture-verified in Dash.

Inspect the audio capability layerASR, TTS, and audio-understanding building blocks. Read the Agent integration boundaryKeep provider credentials server-side and use one versioned workflow source.