Visible token economics
Review published input and output rates. The cost decision stays in your hands, without an unverifiable “lowest price” promise.
Specialized catalog
China-developed AI models, ready to integrate.
Explore DeepSeek, Qwen, Kimi, GLM, and other families through an OpenAI-compatible interface. Compare published input and output costs before sending traffic.
- 10
- available models
- 4
- provider families
Commercially reviewed catalog
Only available routes that pass NexoRouter publication and billing controls appear here.
Traceable operations
Use prepaid balance, separate keys, and usage logs to review consumption, errors, and activity.
Verifiable directory
Compare models by workload and budget.
Prices use the billing unit published in the catalog. Open a profile to review the model ID, endpoint, editorial context, and request example.
Qwen3-30B-A3B
qwen3-30b-a3b
- Input
- $0.105 / 1M tokens
- Output
- $0.42 / 1M tokens
Qwen3.5-Plus
qwen3.5-plus
- Input
- $0.112 / 1M tokens
- Output
- $0.672 / 1M tokens
DeepSeek V3.2 Exp
deepseek-ai/DeepSeek-V3.2-Exp
- Input
- $0.28 / 1M tokens
- Output
- $0.42 / 1M tokens
Qwen3.6-Plus
qwen3.6-plus
- Input
- $0.28 / 1M tokens
- Output
- $1.68 / 1M tokens
Kimi K2.5
moonshotai/kimi-k2.5
- Input
- $0.56 / 1M tokens
- Output
- $2.94 / 1M tokens
Mimo V2.5
mimo-v2.5
- Input
- $1 / 1M tokens
- Output
- $3 / 1M tokens
Mimo V2.5 Pro
mimo-v2.5-pro
- Input
- $1 / 1M tokens
- Output
- $3 / 1M tokens
Kimi K2.6
kimi-k2.6
- Input
- $1.5 / 1M tokens
- Output
- $6.25 / 1M tokens
Kimi K2.7 Code
kimi-k2.7-code
- Input
- $1.5 / 1M tokens
- Output
- $6.25 / 1M tokens
Kimi K3
kimi-k3
- Input
- $4.65 / 1M tokens
- Output
- $23.1 / 1M tokens
What “secure and standardized” means at NexoRouter
We do not use those words as an absolute guarantee. We translate them into product controls a customer can verify.
01
Access
Independent API keys and secrets that are not exposed on public pages.
02
Billing
Prepaid balance, published rates, and a top-up record.
03
Observability
Usage logs, response status, and aggregate reliability signals.
04
Policies
Terms, privacy, and acceptable-use rules linked from every public page.
Try a China-developed model family without rebuilding your integration.
Change the base URL and model ID in an OpenAI-compatible client.