Kimi-K2.6, 9× cheaper
The same open-weight Kimi-K2.6 — fixed USD per-token, 200K context, drop-in via the OpenAI API. Built for reasoning and long-context work.
Kimi-K2.6 is temporarily unavailable. Available right now: GLM-5.3-Flash, DeepSeek-V4-Flash, MiniMax-M2.7.
Why run it here
Published per-million-token USD rates, about 9× below OpenRouter on the same model. Pay-as-you-go from a USD balance — predictable cost for agents and coding.
Drop-in, OpenAI-compatible
Point any OpenAI-compatible client at our base URL and select Kimi-K2.6. The full 200K context and standard tool calls pass straight through.
Use Kimi-K2.6
# Kimi-K2.6 via cheaptokens (OpenAI-compatible)
export OPENAI_BASE_URL="https://api.cheaptokens.dev/v1"
export OPENAI_API_KEY="YOUR_CHEAPTOKENS_KEY"
# model: moonshotai/Kimi-K2.6Kimi-K2.6 vs OpenRouter
Per-million-token rates on the exact same model.
| Model | OpenRouter | cheaptokens | You save |
|---|---|---|---|
| GLM-5.3-Flash temporarily unavailable | $0.048 / $0.20 | $0.027 / $0.08 | ~2× cheaper |
| DeepSeek-V4-Flash temporarily unavailable | $0.016 / $0.64 | $0.0221 / $0.10 | ~4× cheaper |
Same models. USD per 1M tokens (input / output). OpenRouter shown for comparison; savings by blended price (3 parts input to 1 part output). · Open cost calculator
FAQ
Is this the same Kimi-K2.6?
Yes — the same open-weight Kimi-K2.6, served at a fixed USD per-token rate.
Is it good for reasoning and long context?
Yes — Kimi-K2.6 handles the full 200K context and reasoning-style prompts.
How do I pay?
Top up a USD balance with crypto — no KYC — or by card through a partner, and spend per-token.
Related use cases
Start with Kimi-K2.6
Create an account, top up with crypto, and call Kimi-K2.6 at a fixed per-token rate.
Get startedOpen-weight model served on cost-efficient infrastructure. Model names are trademarks of their respective owners.