gemini-3-flash
$0.10 in · $0.40 out
CI-AI Gateway gives you a single OpenAI-compatible endpoint in front of GPT-5, Claude, Gemini, DeepSeek, Qwen, Llama and Kimi. Smart routing, automatic fallbacks, response caching, BYOK and open-weight models you can run on your own GPU.
Create an API key Open the demo chat
OpenAI-compatible /v1/chat/completions · streaming · function calling · GET /v1/models
curl https://ci-ai.duckdns.org/v1/chat/completions \
-H "Authorization: Bearer $CIAI_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-flash-4.1",
"messages": [
{"role": "user",
"content": "Summarize this changelog"}
],
"stream": true
}'
$0.10 in · $0.40 out
Alibaba · open weights
$0.06 in · $0.24 out
DeepSeek · open weights
$0.08 in · $0.32 out
OpenAI
$1.25 in · $10.00 out
Anthropic
$3.00 in · $15.00 out
Alibaba · open weights
$0.10 in · $0.40 out
| gemini-3-flash | 83.6B |
| qwen-flash-next-3.8 open | 71.3B |
| gpt-5-mini | 61.4B |
| deepseek-flash-4.1 open | 58.9B |
| gpt-5 | 48.2B |
Full breakdown on the catalogue page.
30 seconds in the panel, no card required: 100k trial tokens are already on the account. Scope each key by models, budget and region.
Base URL → https://ci-ai.duckdns.org/v1. Any OpenAI-compatible code, SDK or framework keeps working as-is.
model: "auto" or the :fast / :cheap / :eu suffixes — the gateway picks the upstream from live metrics and keeps a fallback chain ready.

Yes — base URL and auth header change, nothing else. /v1/chat/completions, /v1/embeddings, /v1/models, streaming SSE and tool calls follow the OpenAI spec; vendor-native extras are passed through under extra_body.
Never. Prompts are forwarded to the selected upstream under zero-retention agreements and are not logged beyond 24-hour request metadata unless you enable request archiving per key.
Your key keeps working: the router moves to the next model in your fallback chain and stamps the response with an x-gateway-fallback header.
Yes: open-weight catalogue models are mirrored in our repo and run locally (24 GB VRAM at 4-bit). Enable "burst": "gateway" and overflow peaks are served at catalogue prices.
100k trial tokens, no card. One key — forty-one models.