Kimi K2.5 API: pricing, base URL and quick setup

Specifications and pricing

Prices synced with platform billing (USD per 1M tokens).

Quick setup

curl https://a2agent.me/v1/chat/completions \
  -H "content-type: application/json" \
  -H "authorization: Bearer sk-your-a2agent-key" \
  -d '{
    "model": "kimi-k2.5",
    "messages": [{"role": "user", "content": "Reply with OK"}]
  }'

Using it from a coding client

The 256K window handles a package or a long conversation without chunking. In Cline or Roo Code, where every turn resends the context, the lower input rate is the one you feel.

• Run it in Cline
• Run it in Roo Code

How it compares within its family

Kimi K2.5 and K2.6 both run a 256K context and both read images. K2.5 lists at $0.60 per million input tokens and $3.00 output; K2.6 lists at $0.95 and $4.00. That makes K2.5 the volume option in the family, roughly a third cheaper on input for the same window and the same modality. Two other Moonshot models sit here. K2.7 Code keeps the 256K window, carries the coding tag, and is priced level with K2.6. K3 jumps to a 1M window at $3.00 and $15.00, five times the K2.5 input rate. The ladder inside Moonshot is legible: K2.5 for bulk work, K2.6 for the newer model at the same window, K2.7 Code for code, and K3 only when 256K genuinely will not hold the prompt.

Other models on this platform

• Kimi K2.6256K
• Kimi K2.7 Code256K
• Kimi K31M

Questions