OpenAI
Anthropic
Gemini
DeepSeek
Grok
QwenGPT-6 Astra
OpenAI · 1.05M
INPUT / 1M TOKENSList $10.00$0.70
OUTPUT / 1M TOKENSList $50.00$3.50
Unified model API for AI teams
OpenAI and Anthropic compatible, with routing and failover inside the platform.
40+Models online
99.98%Routing availability
2 protocolsOpenAI / Anthropic compatible
No plansPay per usage
Every Frontier Lab, One Endpoint
OpenAIAnthropicGeminiDeepSeekGrokQwen
Transparent pricing, consistently below list
Metered by official tokenizers, discounts always visible, never above list price.
GPT-5.6 Sol
OpenAI · 1.05M
INPUT / 1M TOKENSList $5.00$0.30
OUTPUT / 1M TOKENSList $30.00$1.80
GPT-5.6 Terra
OpenAI · 1.05M
INPUT / 1M TOKENSList $2.00$0.14
OUTPUT / 1M TOKENSList $12.00$0.84
| Model | Context | Input (list) | Input (Unio) | Output (list) | Output (Unio) | Cache read |
|---|---|---|---|---|---|---|
 GPT-6 Astra OpenAI | 1.05M | $10.00 | $0.70 | $50.00 | $3.50 | $0.07 |
| GPT-5.6 Sol OpenAI | 1.05M | $5.00 | $0.30 | $30.00 | $1.80 | $0.03 |
| GPT-5.6 Terra OpenAI | 1.05M | $2.00 | $0.14 | $12.00 | $0.84 | $0.014 |
| GPT-5.5 OpenAI | 1.05M | $5.00 | $0.30 | $30.00 | $1.80 | $0.03 |
No plans, pay for usage
Your balance is kept in USD and works across every model. Each call settles instantly at the live discount, itemized line by line, and the balance never expires.
- Free trial credit on sign-up, pay after it works
- No monthly fee, no minimum spend
- Exportable usage details for expensing and audits
Balance$50.00
Settled this month$127.40
vs. list priceSaved $509.60
gpt-5.6-sol
in 412K · out 96K$0.71claude-sonnet-5
in 1.2M · out 210K$1.69gemini-3.7-flash
in 640K · out 88K$0.12Cache read
2.1M hits · 91.2% hit rate$0.16Point the endpoint at UnioAPI, keep the rest
The SDKs, CLIs, and agent tools you already use connect almost unchanged.
export ANTHROPIC_BASE_URL="https://api.unioapi.com"
export ANTHROPIC_AUTH_TOKEN="<UNIO_API_KEY>"
claudeProtocol AnthropicEst. time About 1 minute
Built for production traffic
Metrics come from real gateway operations, the same numbers you see in the console.
Price · vs. list0%off
Per-model discounts are published, float with the market, and are capped at list price.
TTFT0.0s
On par with direct connections; slow channels are removed automatically.
Routing availability0.00%
Same-model candidate pools switch over automatically without dropping requests.
Three commitments, built into the product
UnioAPI
One Gateway, Every Model
What you request is what runs
Requests route to the exact model you asked for. No silent substitution, no distilled variants, and the routing result is verifiable in response headers.
Data passes through, never stays
Request and response content is used only to complete the call, meter usage, and debug failures. Never for training, never shared with third parties.
Every cent is auditable
Usage detail down to a single request: model, tokens, discount, and amount are all queryable and exportable, with automatic cut-off on budget overrun.
Platform usage overview
Explore token distribution, monthly trends and daily rankings for models currently on sale.
Loading model usage…
Frequently Asked Questions
UnioAPI supports OpenAI-compatible and Anthropic-compatible APIs. Available models depend on your API key and route configuration.