Intelligence
for everyone.

The latest frontier models, free and open source favorites, and specialists worth discovering. An open marketplace for people who use AI and people who power it.

Become a providerExplore models

compatible

OpenCodePiOh My PiOpenAIAnthropicGoogleMetaMistral AIDeepSeekNVIDIAxAIZ.aiQwenMoonshot AICohereMicrosoftAmazonAI21 LabsPerplexityMiniMaxNous ResearchDatabricksAllen Institute for AIBaiduByteDanceOpenRouterPPQ.aiOllamaDeep CogitoEssential AILiquid AIPrime IntellectNex AGIStepFunUpstageInceptionAion LabsInflectionIBM GraniteHugging FaceBlack Forest LabsTencent HunyuanXiaomi MiMoArcee AIPoolsideMeituan LongCatSakana AIRelaceTypeSafeThinking MachinesReka AIinclusionAIKwaipilotWriterApodexPerceptronvLLMllama.cppTogether AINovita AIDeepInfraStreamLakeGMI CloudCoreWeaveWaferAzureParasailOpen InferenceAtlas CloudFireworks AISiliconFlowSail ResearchModalinference.netPhalaVeniceDigitalOceanBasetenNextBit

An open marketplace

Let’s put every GPU to work.
For you.

For people who use AI

Find the model that does it better.

Reach for frontier intelligence, build with free and open source models, or discover specialists that are hard to find anywhere else. Providers compete, so you get the cheapest available offer for what you actually use.

Find your next model

For people who power AI

Put your spare capacity to work.

An idle GPU. Unused hardware. More AI tokens than you need. Turn spare capacity into inference someone else can use — from a single machine to full data centers — and earn when people run your models.

Become a provider

Payments

Your way to pay.

Pay securely by credit card through Stripe, or use Bitcoin. Add funds to one account and share access keys with your team or organization.

Integrates with everything

Keep your harness.

OpenCode, Pi, OMP, and other compatible harnesses — more models in the workflow you already love. One account spans every inference source: frontier APIs, OpenRouter, and independent providers serving Ollama, vLLM, or llama.cpp.

  • OpenCode
  • Pi
  • Oh My Pi
  • Any compatible client
OpenCodePiOh My PiOpenAIAnthropicGoogleMetaMistral AIDeepSeekNVIDIAxAIZ.aiQwenMoonshot AICohereMicrosoftAmazonAI21 LabsPerplexityMiniMaxNous ResearchDatabricksAllen Institute for AIBaiduByteDanceOpenRouterPPQ.aiOllamaDeep CogitoEssential AILiquid AIPrime IntellectNex AGIStepFunUpstageInceptionAion LabsInflectionIBM GraniteHugging FaceBlack Forest LabsTencent HunyuanXiaomi MiMoArcee AIPoolsideMeituan LongCatSakana AIRelaceTypeSafeThinking MachinesReka AIinclusionAIKwaipilotWriterApodexPerceptronvLLMllama.cppTogether AINovita AIDeepInfraStreamLakeGMI CloudCoreWeaveWaferAzureParasailOpen InferenceAtlas CloudFireworks AISiliconFlowSail ResearchModalinference.netPhalaVeniceDigitalOceanBasetenNextBit

Use the world’s best models

You choose the model.
We find the way.

The same model can come from many providers. We compare live price, speed, and reliability and route each request to a source that’s delivering right now — lower cost, quality still in the picture.

Your request

author/model

  1. Source A$0.42 · 0.28 sSelected
  2. Source B$0.55 · 0.21 sFallback
  3. Source C$0.38 · —At capacity
  4. Source D$0.40 · 0.64 sFailing checks
How selection works · illustration, not live provider data.

  1. 01

    Better prices

    Providers compete for your requests. Compare the lowest available prices for the model you want.

  2. 02

    Best speed

    We continually measure response times and available capacity to find sources ready for your next request.

  3. 03

    Reliability first

    Ongoing health checks keep unavailable sources out of the running. If a source fails before answering, we can try another.

Use models.
Or serve them.

Pay per request, or earn from your spare capacity.

Become a providerExplore models