Building summarization and need a China-hosted model that works from abroad? Here is the shortlist and the one-key fix.
For summarization, the priorities are usually bilingual quality, latency to your users, and predictable cost. Chinese-origin models (Doubao, Qwen, GLM, Hunyuan, DeepSeek, Kimi) are priced well below US equivalents and handle Chinese natively.
| Model | Input | Output | Fit for summarization |
|---|---|---|---|
| Doubao | ¥0.6 | ¥3.6 | 32K-context Lite; strong Chinese; via Volcano Ark. |
| Qwen | ¥0.3 | ¥0.6 | Qwen-Turbo; best cost-performance for English+Chinese. |
| GLM | ¥0.6 | ¥0.6 | GLM-4-Air; GLM-4-Flash has a permanent free tier. |
| Hunyuan | ¥0.8 | ¥2 | TurboS; hunyuan-lite is completely free. |
| DeepSeek | ¥1 | ¥2 | V3; available via SiliconFlow aggregation. |
| Kimi | ¥1 | ¥1 | Long-context specialist (200K+). |
curl https://447a931e9b6e46249e80a111415cd170.app.workbuddy.link/v1/llm \
-H "Authorization: Bearer YOUR_KEY" \
-d '{"provider":"auto","tier":"cost","messages":[{"role":"user","content":"Your summarization prompt"}]}'
Use tier:"cost" for cheapest, tier:"quality" for best output. Same price, better uptime via automatic failover.
TideLink wraps Doubao, Qwen, GLM, Hunyuan, DeepSeek and Kimi behind a single OpenAI-compatible endpoint — with free sign-up credit.
Get an API key →