ModelScope
Catatan dari sumber
Free API-Inference for registered users. Requires Alibaba Cloud account binding + real-name verification.
Cara claim API key gratis
Langkah umum — detail pastinya ikutin halaman resmi ModelScope.
- 1.Buka halaman API key ModelScope ↗
- 2.Daftar akun baru, atau login kalau udah punya.
- 3.Generate API key di dashboard / settings.
- 4.Pakai API key + Base URL
https://api-inference.modelscope.cn/v1di SDK atau HTTP client.
Model tersedia (48+)
- Modality
- —
- Context
- 72
- Output
- 8K
- Modality
- reasoning
- Context
- 70
- Output
- 205K
- Modality
- —
- Context
- 69
- Output
- 8K
- Modality
- —
- Context
- 58
- Output
- 8K
- Modality
- —
- Context
- 58
- Output
- 8K
- Modality
- —
- Context
- 52
- Output
- 8K
- Modality
- reasoning
- Context
- 46
- Output
- 200K
- Modality
- reasoning
- Context
- 44
- Output
- 1.0M
| Model | Modality | Context | Output | Rate limit |
|---|---|---|---|---|
deepseek-ai/DeepSeek-V4-Pro deepseek-ai-deepseek-v4-pro | — | 72 | 8K | — |
MiniMax-M2.5-highspeed minimax-m2-5-highspeed | reasoning | 70 | 205K | — |
deepseek-ai/DeepSeek-V4-Flash deepseek-ai-deepseek-v4-flash | — | 69 | 8K | — |
Qwen/Qwen3.5-122B-A10B qwen-qwen3-5-122b-a10b | — | 58 | 8K | — |
Qwen/Qwen3.5-397B-A17B qwen-qwen3-5-397b-a17b | — | 58 | 8K | — |
deepseek-ai/DeepSeek-V3.2 deepseek-ai-deepseek-v3-2 | — | 52 | 8K | — |
GLM-4.7-FlashX glm-4-7-flashx | reasoning | 46 | 200K | — |
GLM-5.2 glm-5-2 | reasoning | 44 | 1.0M | — |
+ API-Inference-enabled models — daftar lengkapnya ada di sumber.
FAQ
Apakah API ModelScope gratis?
Ya — menurut sumber: "Free API-Inference for registered users. Requires Alibaba Cloud account binding + real-name verification." Data dari mnfst/awesome-free-llm-apis, freellm.net, terakhir di-sync 19 Jul 2026 — cek sumber untuk kondisi terbaru.
Ada berapa model gratis di ModelScope?
48 model tercantum per 19 Jul 2026, termasuk deepseek-ai/DeepSeek-V4-Pro, MiniMax-M2.5-highspeed, deepseek-ai/DeepSeek-V4-Flash.
Berapa rate limit ModelScope?
Beda-beda per model, contoh: Qwen/Qwen3.5-27B: 2,000 RPD total; <=500 RPD/model (dynamic); Qwen/Qwen3.5-35B-A3B: 2,000 RPD total; <=500 RPD/model (dynamic). Limit lengkap per model, lihat tabel di atas.
Berapa context window maksimal di ModelScope?
72 (deepseek-ai/DeepSeek-V4-Pro).