OpenAI-compatible gateway routing your requests to the cheapest available provider across Novita, Together, Gonka, and more. Pay only for what you use.
# OpenAI Python SDK with one line changed
from openai import OpenAI
client = OpenAI(
api_key="sk-live-...",
base_url="https://api.cocolada.xyz/v1",
)
r = client.chat.completions.create(
model="llama-3.3-70b",
messages=[{"role": "user", "content": "Hi"}],
)
print(r.choices[0].message.content)Drop-in OpenAI SDK. Send to llama-3.3-70b — we route through Novita, fail over to Together. You change nothing.
Catalog shows the upstream price alongside our 15% markup. No hidden fees, no per-minute rates, no credit-pack gimmicks.
Top up with USDC on Base or Solana. No card walls, no KYC for prepaid usage. Move money in 30 seconds.
Native SSE for streaming, structured outputs, function calling. The full OpenAI API surface, mirror-matched.
Three-tier rate limiting per IP, key, and account. Scoped sub-keys for delegation. Sleep through abuse spikes.
Roadmap: Gonka (verifiable inference) and Cocoon (TEE-confidential). Provable inference when you need an audit trail.
Every model below is served by at least one provider. Same alias, automatic failover, single bill.
See all 190 models →| Model | Context | Input / 1M | Output / 1M |
|---|---|---|---|
Llama 3.3 70B meta-llama/Llama-3.3-70B-Instruct | 131k | $0.135 | $0.400 |
DeepSeek V3.2 deepseek/deepseek-v3.2 | 128k | $0.269 | $0.400 |
Qwen 2.5 72B Qwen/Qwen2.5-72B-Instruct | 32k | $0.300 | $0.300 |
Mixtral 8x22B mistralai/Mixtral-8x22B-Instruct | 64k | $0.880 | $0.880 |
MiniMax M3 minimax/minimax-m3 | 1M | $0.300 | $1.200 |
No subscriptions, no minimum spend. Top up credits when you need them — we add 15% over the upstream rate. The full breakdown is in your dashboard.