Find the right AI model
What will you use it for?
Choose the main type of task you plan to do
Choose a use case to get evidence-based recommendations. How recommendations work
| Model | Company | Links | |||||||
|---|---|---|---|---|---|---|---|---|---|
| 1 | 1.Claude Opus 5 |  🇺🇸Anthropic | 63.1 | 82 | 1M | $0.018 | $5.00 | $25.00 | |
| 2 | 2.Grok 4.6 |  🇺🇸xAI | 60.9 | 57 | 500K | $0.0050 | $2.00 | $6.00 | |
| 3 | 3.GPT-5.6 Sol |  🇺🇸OpenAI | 60.9 | 44 | 1.1M | $0.020 | $5.00 | $30.00 | |
| 4 | 4.Kimi K3 |  🇨🇳Moonshot AI | 59.7 | 65 | 1M | $0.011 | $3.00 | $15.00 | |
| 5 | 5.Qwen3.8 Max |  🇨🇳Alibaba | 58.1 | 39 | 1M | $0.0050 | $2.00 | $6.00 | |
| 6 | 6.Claude Opus 4.8 | 🇺🇸Anthropic | 57.3 | 70 | 1M | $0.018 | $5.00 | $25.00 | |
| 7 | 7.GPT-5.6 Terra | 🇺🇸OpenAI | 56.6 | 104 | 1.1M | $0.0040 | $1.00 | $6.00 | |
| 8 | 8.Grok 4.5 | 🇺🇸xAI | 55.8 | 56 | 500K | $0.0050 | $2.00 | $6.00 | |
| 9 | 9.Claude Sonnet 5 | 🇺🇸Anthropic | 55.3 | 67 | 1M | $0.0070 | $2.00 | $10.00 | |
| 10 | 10.Muse Spark 1.1 |  🇺🇸Meta | 53.2 | 138 | 1M | $0.0034 | $1.25 | $4.25 | |
| 11 | 11.GLM 5.2 |  🇨🇳Z.ai | 52.6 | 140 | 1M | $0.0021 | $0.50 | $3.15 | |
| 12 | 12.GPT-5.6 Luna | 🇺🇸OpenAI | 52.3 | 139 | 1.1M | $0.0004 | $0.10 | $0.60 | |
| 13 | 13.Gemini 3.5 Flash |  🇺🇸Google | 52 | 231.5 | 1M | $0.0060 | $1.50 | $9.00 | |
| 14 | 14.DeepSeek V4 Flash 0731 |  🇨🇳DeepSeek | 51.8 | 153 | 1M | $0.0002 | $0.08 | $0.18 | |
| 15 | 15.Gemini 3.6 Flash | 🇺🇸Google | 51.6 | 202 | 1M | $0.0053 | $1.50 | $7.50 | |
| 16 | 16.Gemini 3.1 Pro Preview | 🇺🇸Google | 46.5 | 101 | 1M | $0.0080 | $2.00 | $12.00 | |
| 17 | 17.MiniMax M3 |  🇨🇳MiniMax | 45.4 | 91 | 1M | $0.0009 | $0.30 | $1.20 | |
| 18 | 18.DeepSeek V4 Pro 0813 | 🇨🇳DeepSeek | 45.3 | 48 | 1M | $0.0009 | $0.43 | $0.87 | |
| 19 | 19.Kimi K2.7 Code | 🇨🇳Moonshot AI | 43 | 118 | 262.1K | $0.0024 | $0.67 | $3.40 | |
| 20 | 20.MiMo-V2.5-Pro |  🇨🇳Xiaomi | 42.9 | 28 | 1.1M | $0.0009 | $0.43 | $0.87 |
Showing 1–20 of 76
1 / 4
Cost per task estimates 1,000 input tokens and 500 output tokens using the listed API prices. Recommendation scores include an uncertainty penalty when relevant benchmark evidence is incomplete.