Payments are live via Creem (Merchant of Record). Serving backend is disclosed per model at /provider/health.
TokenShop

B2B wholesale

Sell tokens to channels

This path sells inference capacity to aggregators (OpenRouter, OneAPI stations, cloud marketplaces) — not end users buying on a retail page. Software readiness is not a live listing until the counterparty accepts.

Live supply status

Owned endpoint
configured
Selected supply mode
local_owned
Owned models
local/qwen3-4b
Externally sellable
local/qwen3-4b
Serving location
CN / local-laptop-rtx5070
Configured limits
24 rpm · 3190/1595 TPM
Listing approved
no (is_ready stays false)
Settlement
prepaid only (human_gate)

Limits are enforcement settings, not proof that hardware is online. Gateway-backed models are test-only and never offered as our own supply.

OpenRouter checklist

  1. Models document (schema 2.4): https://tokenhubapp.site/provider/openrouter/models
  2. Inference: POST https://tokenhubapp.site/v1/chat/completions with Bearer key
  3. Provider health: https://tokenhubapp.site/provider/health
  4. Privacy / Terms (required fields): /privacy · /terms
  5. Apply pack JSON: https://tokenhubapp.site/api/wholesale/apply-pack
  6. Form: openrouter.ai/providers/apply/form — company email and HQ are human decisions
  7. Auto top-up or invoicing after acceptance (human_gate)

Channel board

IDChannelStatusLink
OR1OpenRouter Providerhuman_gate — form prefilled, needs company email + HQopen
E12OneAPI / NewAPI upstreamsoftware-compatible — commercial supply gates still applyopen
E7Direct wholesale quoteindicative API — not a live supply claimopen
A4NanoGPT provider intakehuman_gate — email/Discord intakeopen
H4Zyla / APILayer marketplacehuman_gate — entity + payoutopen
H6Tencent Cloud marketplacehuman_gate — deposit + credentialsopen

Retail leads live at /channels — a different motion from wholesale supply.