Skip to content
Checking session...
<- All providers

Join the network

Register as a Provider

Apply as a company, data center, or professional edge network that can process BatchProviderApi jobs. Approved providers publish model offerings, prices, rate limits, regions, and privacy posture through the same onboarding flow.

Provider identity

Eligibility

Companies and professional networks only

BatchRouter keeps the provider directory small and reputable. We review organizations that operate reliable hosted capacity, data center capacity, or a managed edge network.

  • Accepted: AI companies, data centers, cloud infrastructure operators, and professional edge networks.
  • Not accepted: individual edge nodes, hobby hardware, or one-off machines. Want to join a community edge network? Visit ai.autonomousc.com.
  • Applicants should operate reliable hosted capacity, data center capacity, or a managed edge network.

Public company or provider brand. Individual node names are not accepted.

Use your organization brand. Do not use a machine or node identifier.

Used for review, contracting, settlement, and compliance checks.

Choose edge network only for an organization that manages multiple nodes under one provider contract.

Use a company-owned email address for provider review.

Batch Provider API endpoint

Base URL for your Batch Provider API.

Stored encrypted. Sent as a Bearer token on provider API calls.

Declared capacity & capabilities

Enter the sustained tokens/sec you can keep accepting for normal routed work. This becomes a cold-start routing prior until observed telemetry exists.

Use a conservative prior such as 0.95. Provider profiles show observed telemetry separately, so this is not presented as a measured public stat.

Choose from the shared routing region list. Use Global if capacity is not region-bound.

Preferred chunk size for efficient processing. It is advisory; BatchRouter may split smaller when the job, queue, or SLA requires it.

Hard limit for one assigned provider job. Set this below the item count where your workers or queue become unreliable.

Preferred estimated token volume per chunk. Leave blank when item count is the safer sizing signal.

Hard estimated-token cap for a single chunk. Use it to prevent oversized assignments even when item counts look small.

Parallel chunks you can process safely at the same time. Start conservatively and raise it after successful observed throughput.

Immediate item capacity BatchRouter may reserve for new work. Refresh this later from the portal or heartbeat when capacity changes.

Immediate estimated-token capacity available for queued work. Leave blank if your routing capacity is controlled only by item count.

Optional provider-wide request-rate guardrail for your serving stack or downstream inference provider.

Conservative provider-wide token budget for new offerings. You can replace it with more precise per-offering capacity and heartbeat values in the portal.

Operations supportediSelect every task type your BatchProviderApi endpoint can execute. Responses covers text generation; embeddings covers vector output.

Choose only operations your BatchProviderApi implementation can run end to end, including validation and result submission.

iConfirms this is a registered organization or managed edge network that can be reviewed, contracted, and paid as a provider.

BatchRouter reviews and pays organizations or managed networks, not individual machines.

Choose deny when prompts and outputs are processed transiently only. Choose allow only if your provider terms permit retaining customer payloads.

Reviewed by the BatchRouter operator team.