Coret Inference

An OpenAI-compatible inference API. Prepaid credits, no rate-limit surprises.

Models

Model	Context	Latency (TTFT p50)	TTFT p95	Throughput p50	Uptime 30d
`coret/deepseek-v4-flash` DeepSeek V4 Flash	1024K	—	—	—	—

Latency is time-to-first-token measured at our edge over real traffic; throughput is output tokens per second per request; uptime is the share of 5-minute windows passing a health probe over the trailing 30 days.

Getting started

Point any OpenAI-compatible client at our base URL and use the API key we issued you.

from openai import OpenAI

client = OpenAI(
    base_url="https://inference.coret.ai/v1",
    api_key="sk-coret-...",
)

resp = client.chat.completions.create(
    model="coret/deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello"}],
    stream=True,
)

Streaming, tool calling and JSON mode are supported. Need an account? Talk to us.