Moonshot

Kimi K2.6: Fast

Kimi K2.6 on a dedicated SGLang route with isolated capacity for latency-sensitive agent workloads.

1T MoE (~32B active)Modified MIT

Context

256K

Standard tok/s

TTFT

280ms

Ultra-fast tier serves at 1,000+ tps on reserved capacity, see pricing.

Pricing

Input: $0.68 / 1M tokens
Output: $3.41 / 1M tokens
Cached input: $0.144 / 1M tokens
Fine-tunes: Same per-token price as base

Capabilities

SSE streaming
Tool / function calling
Structured JSON outputs
Reasoning traces

Use cases

Coding agentsLatency-sensitive tool useLong-horizon reasoning

Quickstart

Full quickstart

kimi-k2.6-fast.py

from openai import OpenAI

client = OpenAI(
    base_url="https://api.cogito.decart.ai/v1",
    api_key=os.environ["COGITO_API_KEY"],
)

response = client.chat.completions.create(
    model="moonshotai/kimi-k2.6:fast",
    messages=[{"role": "user", "content": "Hello!"}],
)

print(response.choices[0].message.content)