Long-context chat completions via Telnyx Inference (Kimi K2.6) — 200K-char input window, 2048-token output. OpenAI-compatible. $0.0543 USDC per call. (upto metered — large context)