kimi_k3.py

"""Kimi K3 via NovAI (OpenAI-compatible, drop-in).

Moonshot's Kimi K3: 2.8T-param MoE, native vision, 1M-token context.
Only change vs. the OpenAI SDK: base_url + model="kimi-k3".

    pip install openai
    export NOVAI_API_KEY=sk-...
    python kimi_k3.py
"""
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["NOVAI_API_KEY"],
    base_url="https://aiapi-pro.com/v1",
)

# 1) Plain chat -----------------------------------------------------------
resp = client.chat.completions.create(
    model="kimi-k3",
    messages=[{"role": "user", "content": "Explain mixture-of-experts routing in 3 bullet points."}],
)
print(resp.choices[0].message.content)

# 2) Streaming ------------------------------------------------------------
# for chunk in client.chat.completions.create(
#         model="kimi-k3",
#         messages=[{"role": "user", "content": "Write a haiku about long context."}],
#         stream=True):
#     print(chunk.choices[0].delta.content or "", end="", flush=True)

# 3) Vision (K3 is multimodal) -------------------------------------------
# resp = client.chat.completions.create(
#     model="kimi-k3",
#     messages=[{
#         "role": "user",
#         "content": [
#             {"type": "text", "text": "What's in this image?"},
#             {"type": "image_url",
#              "image_url": {"url": "https://example.com/chart.png"}},  # replace with your image
#         ],
#     }],
# )
# print(resp.choices[0].message.content)

# 4) Long context (up to 1M tokens) --------------------------------------
# with open("big_document.txt") as f:
#     doc = f.read()
# resp = client.chat.completions.create(
#     model="kimi-k3",
#     messages=[{"role": "user", "content": f"Summarize the key risks:\n\n{doc}"}],
# )
# print(resp.choices[0].message.content)