See exactly what you'd pay for GPT-5, Claude, DeepSeek, Qwen โ and how much smart routing saves on a real task mix.
โ 10.0M tokens
โ 3.0M tokens
% of requests by type. Total must equal 100%. Default is typical for chat-heavy AI apps.
Total: 100% โ
Recommended
model="auto" โ picks the cheapest model that can handle each task.
$34.47/mo
Save $55.53 vs always-Claude (62% off)
| Model | Input $/M | Output $/M | Monthly cost |
|---|---|---|---|
| โจGLM-4 Flash | $0.01 | $0.01 | $0.13 |
| ๐Qwen Turbo | $0.08 | $0.31 | $1.73 |
| ๐ซGPT-4o Mini | $0.18 | $0.72 | $3.96 |
| ๐ณDeepSeek V3 | $0.34 | $0.50 | $4.90 |
| ๐ณDeepSeek R1 | $0.34 | $0.50 | $4.90 |
| ๐Kimi K2.5 | $0.24 | $1.20 | $6.00 |
| ๐Qwen Plus | $0.13 | $1.87 | $6.91 |
| โกGemini 2.5 Flash | $0.36 | $3.00 | $12.60 |
| โGemini 2.5 Pro | $1.50 | $12.00 | $51.00 |
| ๐คGPT-5 | $3.00 | $18.00 | $84.00 |
| ๐ง Claude Sonnet 4 | $3.60 | $18.00 | $90.00 |
| ๐ฆClaude Opus 4.6 | $6.00 | $30.00 | $150.00 |
Pricing per 1M tokens (INTL tier โ most users). CN-region pricing on the /pricing page. All prices same as direct provider; AIPower covers gateway / failover / WeChat Pay / billing.
10 free trial calls. $5 first top-up adds +100 bonus calls and unlocks all 16 models including smart routing.