Moonshot Kimi K3 — flagship model with a 1M-token context window, multimodal understanding, and strong fit for long-horizon coding and knowledge work.
Key strengths
- 1M-token context
- Reasoning mode
- Multimodal input
- Long-horizon coding agents
- Tool calling
Use cases
- Repository-scale coding agents
- Deep research
- Multimodal debugging
- End-to-end knowledge work
When Thinking mode is enabled, reasoning tokens generated by the model are counted as billable output. This can increase total usage beyond the visible answer tokens.
Moonshot's moonshot/Kimi-K3 is a frontier text generation model in the Kimi family. It excels at complex reasoning, agentic workflows, code generation, and long-form writing tasks, with native support for streaming, tool calling, JSON mode, and multi-turn conversations.
The model handles long-context inputs gracefully and is particularly effective for software engineering, multi-step research, and end-to-end project execution. Its tokenizer and pricing are optimized for high-throughput production workloads, with a competitive cost profile relative to other models in its tier.
moonshot/Kimi-K3 is fully OpenAI-compatible — drop in your existing OpenAI Python or Node SDK and switch `baseURL` to `https://api.tokenlx.ai`. TokenLX transparently routes your requests to the optimal provider endpoint while preserving streaming, function-calling, and structured-output semantics.
Performance
Compare different providers across TokenLX · All locations.
Effective Pricing
Actual cost per million tokens across providers over the past 7 days.
Recent activity
Total usage per day on TokenLX (last 30 days).
Sample code & API
TokenLX normalizes requests and responses across providers. Use any OpenAI SDK or our native SDK.
Disable Thinking when you do not need explicit reasoning, or set a lower budget_tokens value to cap the reasoning length. Only enable return_thoughts when you need to inspect the thinking process.
from openai import OpenAI
client = OpenAI(
base_url="https://api.tokenlx.ai/v1",
api_key="sk-aihub-...",
)
# Non-streaming
response = client.chat.completions.create(
model="Kimi-K3",
messages=[
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Hello!"},
],
# Optional: enable Thinking / reasoning.
extra_body={
"thinking": {
"enabled": True,
"budget_tokens": 2048,
"return_thoughts": True,
}
},
)
print(response.choices[0].message.content)
# Streaming
stream = client.chat.completions.create(
model="Kimi-K3",
messages=[{"role": "user", "content": "Tell me a story"}],
stream=True,
extra_body={
"thinking": {
"enabled": True,
"budget_tokens": 2048,
"return_thoughts": True,
}
},
)
for chunk in stream:
print(chunk.choices[0].delta.content or "", end="", flush=True)Replace sk-aihubrouter-… with your key from the dashboard.
API Endpoints
Sends a request for a model response for the given chat conversation. Supports both streaming and non-streaming modes.
Creates a streaming or non-streaming response using the OpenAI Responses API format.
Creates a message using the Anthropic Messages API format. Supports text, images, tools, and extended thinking.