Inference
Hosted Inference
Use curated amics models through an OpenAI-compatible API or the SDK.
Hosted inference is the fastest path: choose one of the curated amics model ids and send a chat completion request.
Model IDs
amics exposes only the models enabled for your workspace. The current hosted catalog is curated; provider details are not part of the user-facing API.
Examples:
kimi-k2-7-codeglm-5-2nemotron-ultradeepseek-v4-prokimi-k2-6glm-5-1
Python SDK
from amics import Client
amics = Client()
response = amics.inference.chat(
"deepseek-v4-pro",
[{"role": "user", "content": "Implement Hello World in Python"}],
max_tokens=1000,
temperature=1,
)
print(response["choices"][0]["message"]["content"])Streaming text:
for token in amics.inference.stream_text(
"deepseek-v4-pro",
[{"role": "user", "content": "Write a deployment checklist"}],
):
print(token, end="", flush=True)OpenAI-Compatible Clients
from openai import OpenAI
client = OpenAI(
api_key="amics_sk_...",
base_url="https://ingest.amics.ai/v1",
)
response = client.chat.completions.create(
model="deepseek-v4-pro",
messages=[{"role": "user", "content": "Implement Hello World in Python"}],
stream=True,
)TypeScript SDK
import { Client } from "@amics/sdk"
const amics = new Client()
const response = await amics.inference.chat(
"deepseek-v4-pro",
[{ role: "user", content: "Implement Hello World in TypeScript" }],
{ max_tokens: 1000 },
)
console.log(response.choices)