One OpenAI-compatible key, 300+ models · image2.5 $0.0085/image · Seedance 2.0 Mini $0.01056/sec · LLM from $0.0228 / M tokens · $1 minimum top-up.

See live pricing · Get an API key

openai-api-proxy-2026 keeps one base_url and one key in front of 300+ models — USD settlement, pay-as-you-go, $1 minimum top-up.

curl -X POST https://api.apimart.ai/v1/images/generations \
  -H 'Authorization: Bearer $APIMART_KEY' \
  -H 'Content-Type: application/json' \
  -d '{"model":"gpt-image-2.5-ext","prompt":"cozy reading nook, warm lamp, cinematic","size":"1:1","resolution":"1K","n":1}'

Async: submit → get task_id → poll GET https://api.apimart.ai/v1/tasks/<task_id> → read cost / credits_cost from the result. Parameter tables, version/resolution/size options and idempotency headers are documented on the model page reachable from the pricing link above.

reverse-engineered · reverse engineering · ai api gateway · ai-api-gateway · api relay · 中转站 · nano banana 2 api · gpt-image-2.5 api · ai api pricing · pay-as-you-go

• USD settlement, pay-as-you-go, $1 minimum top-up, no subscription.
• Operating since last year; ~100,000 registered users, mostly enterprise accounts.
• International invoices available on request.
• 307 models online (live /v1/models) as of 2026-09-28.

This repository documents APIMart, a third-party API aggregator/gateway. It is not affiliated with, endorsed by, or sponsored by OpenAI, Google, Anthropic, xAI, ByteDance or any model vendor. Model names and trademarks belong to their owners. Prices are a point-in-time snapshot and may change; the vendor's console billing is authoritative.