OneInfer Voice is live.Try OneInfer Voice

OneInfer Cloud

One API. Every model. The sharpest path.

OneInfer Cloud is our managed AI cloud: one API and one key for text, vision, audio and video models across providers. It works with the OpenAI SDK you already use, so switching takes a new base URL and key.

Every request is routed for speed, cost and resilience: short prompts go to lean, low-latency models, hard questions to frontier models, and if a provider slows down or fails, traffic moves to another one automatically. When you need more, rent GPUs by the hour, give a model its own dedicated capacity, or tune kernels to your workload with Kernel Forge, in preview.

Start buildingSee pricing

Every request takes the sharpest path

Illustration

  1. Request received

  2. Guardrails passed

  3. Routing picks a lean model

  4. Model lean model answers

  5. Response streamed back

Short prompt. Short prompts fly through lean, low-latency models.

Bar lengths show order and relative effort, not measured timings.

Text, images, audio and video through one API

One API and one key cover language, vision, audio and video models. A single flow can look at an image, reason about it and reply with a voice, without stitching separate systems together.

  1. Image input
  2. Reasoning
  3. Voice reply

Kernel Forge gives your hottest path its own engine

Kernel Forge turns slow, generic execution into workload-specific kernels. Our agents study the graph, forge Triton and CUDA candidates, and keep the fastest path for production.

In the Console, Kernel Forge is in preview as SoL Optimizer. See a sample run on the Kernel Forge page.

  1. 1

    Trace hot paths in the model graph

  2. 2

    Generate Triton and CUDA candidates

  3. 3

    Benchmark, compare and ship the winner

Compiler intelligence

Agents inspect tensor shapes, bottlenecks and runtime traces before writing kernels tuned for your exact workload.

Fusion by design

Multiple ops are fused into tighter kernels to reduce memory movement, cut overhead and keep GPUs doing useful work.

Developer first

  1. Step 1

    Install the OpenAI SDK

    OneInfer works with the official OpenAI SDK.

    npm install openai

  2. Step 2

    Point it at OneInfer

    Set the base URL and use your OneInfer API key.

    import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.oneinfer.ai/v1',
  apiKey: process.env.ONEINFER_API_KEY,
});

  3. Step 3

    Call any model

    Send chat completions to any chat model in the catalog.

    const response = await client.chat.completions.create({
  model: 'openai/gpt-4.1',
  messages: [{ role: 'user', content: 'Write a blog post' }],
});

The operating layer for serious AI products

  • Dedicated when you need it

    Move a model onto its own GPU capacity, behind one endpoint, when shared capacity is not enough.

  • Model freedom

    Move between frontier models, open models and specialist endpoints through one consistent interface.

  • Bring your SDK

    Works with the OpenAI SDK you already use. Change the base URL and API key to start calling OneInfer.

  • GPUs on demand

    Rent GPU instances by the hour from the console when you need raw compute.

  • Run it in your own cloud

    OneInfer Private Cloud brings the same stack into your own cloud account or data center. It is in development and set up with our engineers.

  • Pricing you can read

    Clear, usage-based billing that shows unit economics before surprises reach the invoice.

Everything on OneInfer Cloud

Connect from your assistant

Use OneInfer from Smithery, Claude or the MCP Registry. Sign in with your own account.

SmitheryClaudeMCP RegistryAll MCP options

Start building on OneInfer Cloud

One key, every model, and the routing, GPUs and kernels behind them.

Start buildingRun it in your own cloud