One API. Every model. Text, image, video.

Build with every AI model. Skip the integration tax.

Godance is an OpenAI-compatible API that routes to 15+ providers — OpenAI, Anthropic, Sora, Claude, Flux, Gemini, Volcengine Ark, and more. Drop-in for chat; image and video run as async jobs. One credit ledger, one billing line, one SDK.

Start free — no card needed Read the docs

Pass-through pricing + a published platform fee. Card, ACH, SEPA, or USDT. No subscription.

Routes to providers including

  • OpenAI
  • Anthropic
  • Google Vertex
  • Sora
  • Black Forest Labs
  • Volcengine Ark
  • DeepSeek
  • Moonshot
  • Qwen
  • Zhipu
  • MiniMax
  • Baidu
  • xAI
  • Kling
  • Runway
  • Stability

Stop maintaining six SDKs.

Six provider integrations, six rate-limit dashboards, six invoices, six SDK upgrades per quarter. Godance is the one URL that replaces all of them.

Multimodal from day one

Text completions, image generation, video clips, TTS, STT — same auth, same billing, same shape. Switching from a Claude chat to a Sora render is one model name away.

Drop-in for OpenAI chat clients

Point the OpenAI SDK at api.godance.ai/v1 and chat just works: same endpoints, same JSON, same streaming format. Migrating chat is a one-line diff. Image and video take OpenAI's request shape but return a job_id to poll. Errors come back as RFC 7807 problem details.

Pay how your customers pay

Card, ACH, and SEPA via Stripe, or USDT via NOWPayments. One credit ledger, one balance — top it up once and spend it across every model.

Fallback chains, not pager duty

Configure per-model fallbacks. Primary 5xx or times out? Godance retries the next option, transparently. Response carries fallback_used:true so your observability stays honest.

Two regions, one DNS

DigitalOcean primary, Alibaba Cloud secondary. Cloudflare DNS health-checked failover. Per-region provider allowlist enforced in code — your Western-frontier calls never route through Alicloud.

Streaming + signed webhooks

SSE for chat completions. Async jobs for images and video, with Stripe-format HMAC-signed callbacks and a 24-hour rotation grace window. Production-grade DX with none of the fiddly bits.

Less migration, more shipping.

Existing OpenAI SDK code — Python, TypeScript, Java, Go — works without modification. Change one URL. Start calling Claude, Gemini, or any other model the next request.

Same idiom for /v1/models. Image and video generation take the same request shape but run as async jobs — poll /v1/jobs/{id} or get a webhook. 

from openai import OpenAI

client = OpenAI(
  base_url="https://api.godance.ai/v1",
  api_key="gd_live_..."
)

# Same SDK. Different model. No code change.
r = client.chat.completions.create(
  model="claude-3-5-sonnet-20241022",
  messages=[{"role": "user", "content": "Hi"}]
)

Questions, briefly answered

Will my existing OpenAI SDK code work?
For chat, yes — swap base_url to https://api.godance.ai/v1 and use a Godance API key. /v1/chat/completions (streaming + non-streaming) and /v1/models match OpenAI's contract. /v1/images/generations takes OpenAI's request shape but runs async: it returns 202 with a job_id, and you fetch the result from GET /v1/jobs/{id} or receive it at a webhook_url. Video generation works the same way.
Which providers and models do you support?
OpenAI (GPT, Sora, gpt-image), Anthropic (Claude), Google Vertex (Gemini, Imagen, Veo), Black Forest Labs (Flux), Volcengine Ark (Doubao, Seedream, Seedance), DeepSeek, Moonshot (Kimi), Alibaba (Qwen), Zhipu (GLM), MiniMax, Baidu (ERNIE), xAI (Grok), Kling, Runway, and Stability. New providers ship as needed — email us.
What does it actually cost?
Provider list price plus a platform fee: 5% on text, 12% on images, and 10–15% on video depending on the model. Every model's fee is listed at godance.ai/models. No tiers, no seats, no minimums. Top up the credit ledger; spend it across every model.
What happens when a provider goes down?
Configure a fallback chain per model. When the primary returns a 5xx or times out, Godance transparently retries the next option in the chain — the response carries fallback_used:true so you know what happened.
How do I verify webhook deliveries?
Outbound webhooks carry X-Godance-Signature in Stripe format (t=<unix>,v1=<hex>). Per-user signing secret retrievable at GET /v1/webhooks/secret. After rotation, both old and new signatures are accepted for 24 hours so deployments don't drop events.

15+ providers. One bill. Ten minutes to first call.

Free to sign up. Free to read the docs. Pay only for what you generate.

Create an account See pricing