Kimi K3 and GLM 5.3 get cheaper on RouterPlus via Wafer

Wafer's prices for Kimi K3 and GLM 5.3 dropped in the RouterPlus catalog on November 4, 2026. Wafer is now the cheapest input route for both models, and the cheapest output route for GLM 5.3.

What changed

• Kimi K3 on Wafer: input fell from $2.00 to $0.60 per million tokens, and cached input from $0.25 to $0.20. Output stays at $15.00.
• The lowest Kimi K3 input price in the catalog fell 60%, from $1.50 (Sail Research) to $0.60 (Wafer). The lowest output price is still $9.50, on Chutes.
• GLM 5.3 on Wafer: input fell from $0.30 to $0.20 and output from $3.00 to $1.80 per million tokens. Cached input stays at $0.18.
• The lowest GLM 5.3 output price in the catalog fell 28%, from $2.50 (DeepInfra) to $1.80 (Wafer).

Kimi K3 prices on RouterPlus

RouterPlus catalog, November 4, 2026, per million tokens:

GLM 5.3 prices on RouterPlus

What it means

For Kimi K3, the cheapest route now depends on your token mix. Wafer is cheaper than Chutes when a request reads more than about 4.4 input tokens per output token, as in long prompts and agent loops, before caching. Chutes stays cheapest for output-heavy work. For GLM 5.3, Wafer is now lowest on both input and output.

See the Kimi K3 model page and the GLM 5.3 model page.

Sources

1. RouterPlus model catalog (November 4, 2026)
2. RouterPlus: Kimi K3 model page
3. RouterPlus: GLM 5.3 model page