Perbandingan Harga Model AI
Semua model, semua varian, dalam satu tabel. Klik header buat sort, cari yang paling cocok dengan budget. Biaya bulanan dihitung dari skenario sedang (50K input + 20K output Token sehari × 30 hari).
📋 Perbandingan Harga Lengkap
| Model | Varian | Input ($/M) | Output ($/M) | Biaya Bulanan |
|---|---|---|---|---|
| 🧪 Zhipu GLM | GLM-4-Flash Gratis | Gratis | Gratis | Gratis |
| 🦙 Llama | Self-hosted Gratis | Gratis | Gratis | Gratis |
| 🔬 DeepSeek | V3.2 (Cache Hit) Gratis | $0.028 | $0.42 | $0.29 |
| ☁️ Tongyi Qwen | Qwen3.5-Flash Gratis | $0.028 | $0.28 | $0.21 |
| 🫘 Doubao | 1.5 Lite Gratis | $0.042 | $0.083 | $0.11 |
| ⚡ MiniMax | abab6.5 Gratis | $0.069 | $0.14 | $0.19 |
| 💎 Gemini | 2.5 Flash-Lite Gratis | $0.1 | $0.4 | $0.39 |
| ☁️ Tongyi Qwen | Qwen3.5-Plus Gratis | $0.11 | $0.67 | $0.57 |
| 🫘 Doubao | 1.5 Pro Gratis | $0.11 | $0.28 | $0.33 |
| 🦙 Llama | Llama 4 Scout (API) Gratis | $0.12 | $0.35 | $0.39 |
| 🌙 Kimi | K1.5 Gratis | $0.14 | $0.56 | $0.55 |
| ⚡ MiniMax | Text-01 Gratis | $0.14 | $1.39 | $1.04 |
| 🤖 GPT | GPT-4o-mini Gratis | $0.15 | $0.6 | $0.58 |
| 🦙 Llama | Llama 4 Maverick (API) Gratis | $0.2 | $0.6 | $0.66 |
| 🔬 DeepSeek | V3.2 (Cache Miss) Gratis | $0.28 | $0.42 | $0.67 |
| 🌙 Kimi | K2 Gratis | $0.28 | $0.83 | $0.92 |
| 💎 Gemini | 2.5 Flash Gratis | $0.3 | $2.5 | $1.95 |
| ☁️ Tongyi Qwen | Qwen3-Max Gratis | $0.35 | $1.4 | $1.36 |
| 🧠 Claude | Haiku 4.5 Gratis | $1.0 | $5.0 | $4.50 |
| 🤖 GPT | o4-mini Gratis | $1.1 | $4.4 | $4.29 |
| 💎 Gemini | 2.5 Pro Gratis | $1.25 | $10.0 | $7.88 |
| 🤖 GPT | o3 Gratis | $2.0 | $8.0 | $7.80 |
| 🤖 GPT | GPT-4o Gratis | $2.5 | $10.0 | $9.75 |
| 🧠 Claude | Sonnet 4.6 Gratis | $3.0 | $15.0 | $13.50 |
| 🧠 Claude | Opus 4.6 Gratis | $5.0 | $25.0 | $22.50 |
| 🧪 Zhipu GLM | GLM-4-Plus Gratis | $6.94 | $6.94 | $14.57 |
💡 Biaya bulanan dihitung dari: 50.000 input Token + 20.000 output Token sehari × 30 hari
🔍 Temuan Perbandingan
DeepSeek V3.2 (Cache Hit) input cuma $0,028/M, hampir 100x lebih murah dari GPT-4o. Kalo prompt-nya gak banyak berubah dan cache hit rate tinggi, DeepSeek emang super murah.
GLM-4-Flash dan Llama self-hosted sepenuhnya gratis. Flash bagus buat yang ringan, Llama self-hosted cocok buat tim yang punya GPU.
Claude Opus 4.6 output $25/M, pilihan paling mahal. Tapi kualitasnya juga top tier, cocok buat pekerjaan yang quality-critical.
GPT-4o-mini ($0,15/$0,6) dan Doubao 1.5 Lite (¥0,3/¥0,6) paling bagus di kategori "murah tapi cukup bagus".