Artifact Virtual

One endpoint for every model.

Swap your base URL, pick a route, and each request goes to the best live model, with fallbacks and rotated keys behind it.

curl https://your-domain.com/api/v1/chat/completions \
  -H "Authorization: Bearer $ARTIFACT_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"artifact/router","messages":[{"role":"user","content":"Hello"}]}'

Routes

artifact/router

Reads each request and picks the route that fits it.

0 live models
artifact/frontier

The strongest live models, for hard reasoning.

0 live models
artifact/coding

Models tuned for code and long refactors.

0 live models
artifact/agentic

Reliable tool callers for multi-step agents.

0 live models
artifact/fast

Low latency and low cost for everyday chat.

0 live models
artifact/free

No credits needed. Runs on free-tier providers.

0 live models
artifact/image

Text to image. Billed per image.

0 live models
artifact/video

Text to video as a job you poll. Billed per clip.

0 live models
artifact/open

Open-weight models across many GPU hosts, with automatic failover.

0 live models
openaianthropicgooglexaideepseekmistralmoonshotgroqcerebrastogetherfireworksnvidiaopenrouterfaldeepinfrahyperbolicnebiusnovitalambdabasetensambanovahuggingfacecloudflaregithubwandbscalewayovhcloudvultrdigitaloceanparasailchutesfeatherlesscrusoerunpodazure

GPU providers

Routed now

Token-priced, OpenAI-compatible. Add a key and each one goes live.

togetherfireworksdeepinfrahyperbolicnebiusnovitalambdabasetensambanovahuggingfacecloudflaregithubwandbscalewayovhcloudvultrdigitaloceanparasailchutesfeatherlesscrusoegroqcerebrasnvidiaopenrouterrunpodazure

Bring your own endpoint

Rent the GPU, run vLLM or SGLang, then register it in CUSTOM_PROVIDERS.

ModalRunPod podsVast.aiLambda CloudCoreWeavePaperspaceTensorDockHyperstackGenesis CloudFluidStackVoltage ParkVerdaLatitude.shKoyebNorthflankCerebriumBeamSalad

Hyperscalers

Signed or short-lived auth. Put a gateway in front (LiteLLM, Bedrock Access Gateway) and register it.

AWS BedrockAWS SageMakerGoogle Vertex AIGoogle Compute EngineOracle Cloud (OCI)IBM watsonxAlibaba CloudTencent CloudDatabricksSnowflake Cortex

Needs an adapter

Own request format. Not wired yet.

ReplicateStabilityRunwayLumaBlack Forest Labs

Not routable

Interactive quota GPUs with no serving API. Fine for experiments, not for routing.

KaggleGoogle ColabLightning AI StudiosSageMaker Studio Lab
Artifact Router