Solutions
Cheap Chinese AI model API
For DeepSeek, Kimi, GLM, Qwen and MiMo from one key, Somnus gives you 6 models on one OpenAI-compatible key. Usage is billed at list price but every top-up is credited 2× to 2.5×, so DeepSeek V4 Flash — the cheapest option here — effectively costs $0.11 / $0.33 per 1M input/output tokens.
Choosing a cheap chinese ai model api
Chinese labs ship some of the best price-to-quality models available, especially for coding and long context. Signing up with each lab separately means several accounts, payment methods and balances.
Somnus puts them behind one key and one balance, billed in USD, alongside Claude, GPT and Gemini, so you can compare them on your own prompts.
Cheap Chinese AI model API: price ranking
| # | Model | Maker | List (in / out) | Effective with Somnus |
|---|---|---|---|---|
| 1 | DeepSeek V4 Flash | DeepSeek | $0.22 / $0.66 | $0.11 / $0.33 |
| 2 | MiMo V2.5 Pro | Xiaomi | $0.44 / $0.87 | $0.22 / $0.44 |
| 3 | DeepSeek V4 Pro | DeepSeek | $0.66 / $1.98 | $0.33 / $0.99 |
| 4 | GLM 5.3 | Zhipu | $1.4 / $4.4 | $0.7 / $2.2 |
| 5 | Qwen 3.8 Max | Alibaba | $1.65 / $4.95 | $0.83 / $2.48 |
| 6 | Kimi K3 | Moonshot | $3 / $15 | $1.5 / $7.5 |
Ranked by blended price (3 input : 1 output tokens).
Why Somnus for cheap chinese ai model api
- Somnus buys model access in bulk and passes the saving on as extra credit.
- Pay-as-you-go. No monthly caps or throttling — spend your balance however you like.
- $5 free credit on sign-up, no card needed.
- Switch models by changing one string — same key, same balance.
Point the OpenAI SDK at Somnus
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.SOMNUS_API_KEY,
baseURL: "https://gateway-production-c837.up.railway.app/v1",
});
const resp = await client.chat.completions.create({
model: "claude-sonnet-5",
messages: [
{"role": "user", "content": "Cheap Chinese AI model API"}
],
});
console.log(resp.choices[0].message.content);Pick the model name from your dashboard.
| Claude Sonnet 5 | GPT-5.6 Sol | |
|---|---|---|
| Input / 1M | $2 | $5 |
| Output / 1M | $10 | $30 |
| This request | $0.0120 | $0.0340 |
| Blended / 1M (3:1) | $4.00 | $11.25 |
Claude Sonnet 5 is 2.8× cheaper for this request.
Questions
What is the cheap chinese ai model api?
Among DeepSeek, Kimi, GLM, Qwen and MiMo from one key, DeepSeek V4 Flash has the lowest list price on Somnus ($0.22 / $0.66), about $0.11 / $0.33 effective.
Is it really that cheap?
Yes: usage is charged at list price, and each $1 you top up becomes $2–$2.50 of credit.
Related
Browse more: Alternatives · Compare · Use cases · Developers · Solutions · Tools