Transparent billing

Pay for the requests you make.

Current pay-as-you-go pricing for seven Chinese frontier models, with provider-list comparisons, token rates and a clear 20% discount.

Rates are verified against the current gateway mapping. PayPal wallet top-ups are available after sign-in.

Current rates are 20% below the referenced provider list prices. All amounts are USD per 1M tokens.

Verified 2026-08-06 06:03 UTC
DeepSeekDeepSeek V4 Flashdeepseek-v4-flash
Input / 1M$0.112$0.140
Output / 1M$0.224$0.280Cache $0.00224
20% off
DeepSeekDeepSeek V4 Prodeepseek-v4-pro
Input / 1M$0.348$0.435
Output / 1M$0.696$0.870Cache $0.00290
20% off
GLMGLM 5.2glm-5.2
Input / 1M$0.877$1.096
Output / 1M$3.068$3.836Cache $0.219
20% off
Moonshot AIKimi K2.7 Codekimi-k2.7-code
Input / 1M$0.760$0.950
Output / 1M$3.200$4.000Cache $0.152
20% off
Moonshot AIKimi K3kimi-k3
Input / 1M$2.400$3.000
Output / 1M$12.000$15.000Cache $0.240
20% off
MiniMaxMiniMax M3minimax-m3
Input / 1M$0.460$0.575
Output / 1M$1.841$2.301Cache $0.092
20% off
Alibaba CloudQwen 3.8 Maxqwen3.8-max
Input / 1M$1.315$1.644
Output / 1M$3.945$4.932Cache rate not listed
20% off

Pricing evidence

Visible source. Visible calculation.

Method
Provider list price × 0.80. CNY list rates use the gateway's 7.3 CNY/USD conversion snapshot.
Verified
2026-08-06 06:03 UTC · reviewed by Zoryntel operations
Sources
DeepSeek · GLM · Kimi K2.7 · Kimi K3 · MiniMax · Qwen
Change history
2026-08-06 — seven-model pay-as-you-go pricing activated.

PayPal Live wallet top-ups are enabled. The final charged amount is shown before checkout. No recurring or automatic renewal is represented on this page.

How billing works

Separate the numbers developers actually need.

Input tokens

Tokens sent to the upstream that are not eligible for a cache-hit rate.

Output tokens

Tokens generated by the model. Long agent loops can make this the dominant cost.

Cache-hit tokens

Eligible repeated prompt content billed at the published cache-hit rate when the upstream reports it.

PAYG versus a coding plan

Cheapest depends on how consistently you use it.

A fixed coding plan may be cheaper for a heavy user who consumes the full allowance every month. Zoryntel is aimed at irregular usage, multiple clients, programmatic access and teams that need request-level visibility.