FriendliAI (FriendliAI)
综合排名 # ⭐ 可考虑
⚠️ Partially available — servers in US & Korea, ~200-400ms latency from CN | 🌍 国际
💰 Token 价格
| 类型 | 价格 | 备注 |
|---|---|---|
| 输入 (Input) | GLM-5.2/5.1: $1.4/M, DeepSeek-V3.2: $0.5/M, Qwen3-235B: $0.2/M, MiniMax-M2.5: $0.3/M, Gemma-4-31B: $0.14/M | 每百万 tokens |
| 输出 (Output) | GLM-5.2/5.1: $4.4/M, DeepSeek-V3.2: $1.5/M, Qwen3-235B: $0.8/M, MiniMax-M2.5: $1.2/M, Gemma-4-31B: $0.4/M | 每百万 tokens |
💡 免费额度
No free tier — per-token billing with no minimum
🤖 支持模型（共 8 个）
GLM-5.2 GLM-5.1 DeepSeek-V3.2 Qwen3-235B-A22B-2507 MiniMax-M2.5 Gemma-4-31B-it K-EXAONE-236B Whisper-large-v3
✨ 优势
- ✓ 593K+ open models supported via dedicated endpoints
- ✓ SOC 2 Type II & HIPAA compliant
- ✓ Pay-per-token Model APIs + per-second Dedicated Endpoints
- ✓ Prompt caching (cached reads at 0.1x-0.5x input cost)
⚠️ 不足
- × Korean provider — smaller ecosystem vs US competitors
- × No free tier for Model APIs
- × Fewer enterprise integrations (LangChain, etc.)
🎯 适合场景
开发者和团队需要前沿推理 API 或专属 GPU 端点，尤其亚洲用户