One API for every
frontier model.
Call 126 models through a single endpoint. Pay per token from one balance — no subscription.
api.ultragpt.pro/v1- OpenAI-compatible
- Anthropic-compatible
- Cost on every response
UltraGPTcurl https://api.ultragpt.pro/v1/chat/completions \
-H "Authorization: Bearer $ULTRAGPT_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "grok-4.7",
"messages": [{ "role": "user", "content": "Hello!" }]
}'usage: 21 tokens"cost": 0.0000953- models
- 126models
- AI labs
- 11AI labs
- from, per 1M input tokens
- $0.019from, per 1M input tokens
- max context window
- 1.1Mmax context window
Models from 11 leading AI labs
Models
The best model for every job
Switch models by changing one string. Transparent per-token prices, the same as you see on your invoice.




Usage packages
Prepay and get +100% credit
Buy a package and get +100% extra credit for 36 top models: $100 buys $200 of usage. From 1 day to 3 months, from $10. Cached input tokens are not discounted.
60-day Pro
Valid 60 days
- +100% credit on 36 top models
- About $3.33 of usage a day over 60 days
- No daily cap, use it at your pace
- Pay with balance, crypto or card
60-day Business
Valid 60 days
- +100% credit on 36 top models
- About $16.67 of usage a day over 60 days
- No daily cap, use it at your pace
- Pay with balance, crypto or card
60-day Enterprise
Valid 60 days
- +100% credit on 36 top models
- About $33.33 of usage a day over 60 days
- No daily cap, use it at your pace
- Pay with balance, crypto or card
Drop-in
Switch in two lines
Keep the SDKs, agents and frameworks you already use. Point them at UltraGPT and every model becomes available behind the same interface.
client = OpenAI(- api_key=os.environ["OPENAI_API_KEY"],+ base_url="https://api.ultragpt.pro/v1",+ api_key=os.environ["ULTRAGPT_API_KEY"],)export ANTHROPIC_BASE_URL="https://api.ultragpt.pro"
export ANTHROPIC_AUTH_TOKEN="$ULTRAGPT_API_KEY"
export ANTHROPIC_MODEL="claude-opus-4.6"
export ANTHROPIC_DEFAULT_OPUS_MODEL="claude-opus-4.6"
export ANTHROPIC_DEFAULT_SONNET_MODEL="claude-opus-4.6"
export ANTHROPIC_DEFAULT_HAIKU_MODEL="claude-haiku-5.5"
claudePlatform
Everything you need to ship to production
OpenAI & Anthropic compatible
Chat Completions, Responses and the Anthropic Messages API. Works with the official SDKs, Claude Code, Cursor, LangChain and the Vercel AI SDK.
Every model, one key
GPT, Claude, Gemini, Grok, DeepSeek and more — plus embeddings, image generation, speech and transcription.
Fallbacks & web search
Fail over to backup models, route by price or latency, and ground answers in live web results with citations.
Pay per token
Prepaid credits shared with the UltraGPT app. Exact cost on every response, a Batch API with no rate limits, and bring-your-own-key.
Production guardrails
Spend caps, daily limits, IP and domain restrictions, read-only keys, key rotation, idempotent retries and webhooks.
Built for teams
Organizations with a shared balance, roles and per-member keys, plus request-level activity, CSV exports and a live status page.
Integrations
Works with the tools you already use
Coding agents, chat apps, frameworks and automation. Step-by-step setup for 42 tools.
OpenAI API Key: <your UltraGPT key>
Override Base URL: https://api.ultragpt.pro/v1
Custom model: claude-opus-4.6Images & speech
One prompt. Every style.
Images, speech and video sit behind the same key and balance as chat. Change the model name to change the result.
“A house standing alone beside the ocean at night.”
/v1/images/generations
/v1/audio/speech“I read your draft twice. The idea is strong — but the opening hides it. Lead with the moment everything changed.”
API surface
Text, images, audio and video. One base URL.
Chat Completions
/v1/chat/completionsOpenAI format, streaming, tools, vision
Messages
/v1/messagesAnthropic format for Claude Code & SDKs
Responses
/v1/responsesStored, stateful conversations
Embeddings & rerank
/v1/embeddingsSearch and retrieval building blocks
Images
/v1/images/generationsText-to-image from leading labs
Speech & transcription
/v1/audio/*Text-to-speech, speech-to-text
Video
/v1/videosText- and image-to-video jobs
Realtime
/v1/realtimeLow-latency voice over WebSocket
Batch
/v1/batchesLarge async jobs, no rate limits

Get started
Up and running in a minute
- 1
Sign in
Use your UltraGPT account — the same balance works in the app and the API.
- 2
Create a key
Scope it to models, IPs or domains and set a monthly spend cap.
- 3
Change the base URL
Point any OpenAI or Anthropic client at UltraGPT. No other code changes.
Pay only for the tokens you use
No subscription and no minimums. The same limits for everyone, no tiers to climb.
Every key gets 600 requests/min and 50 concurrent requests. No usage tiers — pay as you go, set your own daily and per-key spend caps.
- Uptime, 24h
- 0.00%
- Requests, 24h
- 3
Per-model uptime, latency and throughput measured from real API traffic, refreshed every minute.
Is it really a drop-in replacement for OpenAI?
Yes. Point the official OpenAI SDK (or any OpenAI-compatible client) at our base URL and use your UltraGPT key. Anthropic clients work too, through /v1/messages.
How is usage priced?
Per token, at the price listed for each model on the models page. Every response includes its exact cost in usage.cost, and the same number shows up in your activity log.
Do I need a subscription?
No. You top up a prepaid credit balance with crypto or a card, and requests are billed from it. Credits never expire and are shared with the UltraGPT app.
What are the rate limits?
Every live key gets 600 requests per minute and 50 concurrent requests. There are no usage tiers to climb. You can set lower limits per key, plus daily and per-key spend caps.
Can I use it with Claude Code or Cursor?
Yes. Set ANTHROPIC_BASE_URL for Claude Code, or add a custom OpenAI base URL in Cursor, Cline, Continue and similar tools. The SDKs & tools docs have copy-paste setups.
Do you store my prompts?
Not by default. Prompt logging is opt-in per account. When it is on, prompts and outputs are stored encrypted for debugging and deleted after the retention period you choose.

Start building today
One key, one balance, every model. Your first request is a copy-paste away.




