Model AI terbaik berdasar nilai
Peringkat model berdasar inteligensi per dolar. Diperbarui Agustus 2026. Berdasar data benchmark independen.
| # | Model | Penyedia | Skor | Kecepatan | Harga (output) |
|---|---|---|---|---|---|
| 1 | MiMo V2.5 Pro | Xiaomi | 100.0 | 65 t/s | $0.87/M |
| 2 | DeepSeek V4 Flash | DeepSeek | 100.0 | 113 t/s | $0.28/M |
| 3 | Hunyuan Hy3 | Tencent | 99.9 | 69 t/s | $0.56/M |
| 4 | GPT-5.6 Luna | OpenAI | 99.8 | 175 t/s | $1.2/M |
| 5 | DeepSeek V4 Pro | DeepSeek | 99.8 | 66 t/s | $0.87/M |
| 6 | Inkling Small | Thinking Machines | 99.6 | 130 t/s | $1.2/M |
| 7 | Gemini 3.5 Flash-Lite | 99.4 | 389 t/s | $2.5/M | |
| 8 | MiniMax M3 | MiniMax | 99.0 | 93 t/s | $1.2/M |
| 9 | Kimi K2.7 Code | Moonshot AI | 98.3 | 39 t/s | $4/M |
| 10 | HyperNova 60B | Multiverse | 98.2 | 367 t/s | $0.14/M |
| 11 | Qwen3.7 Plus | Alibaba | 98.2 | 54 t/s | $1.6/M |
| 12 | Qwen3.5 122B-A10B | Alibaba | 97.9 | 126 t/s | $2/M |
| 13 | Muse Spark 1.1 | Meta | 97.6 | 208 t/s | $4.25/M |
| 14 | Grok 4.3 | xAI | 97.6 | 129 t/s | $2.5/M |
| 15 | Gemini 3.1 Pro | 97.2 | 136 t/s | $12/M | |
| 16 | Grok 4.5 | xAI | 97.0 | 59 t/s | $6/M |
| 17 | Qwen3.6 Plus | Alibaba | 97.0 | 55 t/s | $3/M |
| 18 | Qwen3.5 397B-A17B | Alibaba | 97.0 | 71 t/s | $2/M |
| 19 | Nemotron 3 Ultra 550B | NVIDIA | 96.8 | 142 t/s | $2.75/M |
| 20 | GPT-5.6 Terra | OpenAI | 95.6 | 127 t/s | $12/M |
MiMo V2.5 Pro
DeepSeek V4 Flash
Hunyuan Hy3
GPT-5.6 Luna
Gemini 3.5 Flash-Lite
MiniMax M3
Kimi K2.7 Code
Qwen3.7 Plus
Muse Spark 1.1
Grok 4.3
Nemotron 3 Ultra 550BTop 5 \u2014 Perbandingan cepat
Metodologi
Peringkat berdasar evaluasi benchmark independen. Skor dihitung dari beberapa benchmark terstandar dan dinormalisasi ke 0–100.