LLM Opex

LLM hosting inside Hong Kong

Which clouds run language models inside Hong Kong, which models each sells there per token, and at what price. A price here is for a region in the country: data sent to the model does not leave it.

Clouds with a region in the country

No region in the country: Anthropic, Cloudflare, Core42, Crusoe, Databricks, DeepInfra, Fireworks AI, Gemini API, Groq, HF Endpoints, Huawei Cloud, IBM watsonx, Mistral, Nebius, OCI, OpenAI, Snowflake, Telnyx, Together AI, xAI.

Models sold in the country, per token

US dollars per million tokens, input and output.
ModelAlibaba CloudAWSAzureGoogle Cloud
GLM 5.2$1.10 in / $3.85 out———
GLM 5.3$1.13 in / $3.96 out———
GPT-5.1——$1.50 in / $12.00 out—
GPT-5.2——$2.10 in / $16.80 out—
GPT-5.4——$3.00 in / $18.00 out—
GPT-5.4 mini——$0.90 in / $5.40 out—
GPT-5.4 nano——$0.24 in / $1.50 out—
GPT-5.5——$6.00 in / $36.00 out—
Kimi K2.7 Code$0.894 in / $3.71 out———
Kimi K3$2.83 in / $14.13 out$3.00 in / $15.00 out——
Llama 3.3 70B——$0.71 in / $0.71 out—
Llama 4 Maverick 17B——$0.25 in / $1.00 out—

List prices collected 4 November 2026, refreshed every three hours. No negotiated rates or taxes.