Category ranking · Cost efficiency
Best Value LLMs
For high-volume applications, the smartest model isn't always the best buy. This ranking balances capability against API price to surface models that deliver the most per dollar — including several open-weight options you can self-host.
Value scores factor in both benchmark strength and token price; they do not include hosting costs for self-deployed models. Data as of 2026-11-20.
| # | Model | Value | Overall | Intelligence | Coding | Price (in/out) | License |
|---|---|---|---|---|---|---|---|
| 1 | DeepSeek · CN | 99 | 83.6 | 82 | 88 | $0.15 / $0.60 | Open source |
| 2 | ByteDance · CN | 94 | 82.7 | 81 | 78 | $0.30 / $1.20 | Proprietary |
| 3 | OpenAI · US | 92 | 89.3 | 90 | 88 | $2 / $10 | Proprietary |
| 4 | Anthropic · US | 90 | 88.9 | 89 | 92 | $2 / $10 | Proprietary |
| 5 | Z.ai (Zhipu) · CN | 90 | 84.5 | 84 | 89 | $0.60 / $2.20 | Open weights |
| 6 | Meta · US | 90 | 78.5 | 77 | 76 | $0.20 / $0.60 | Open source |
| 7 | Alibaba Cloud · CN | 88 | 88.8 | 88 | 86 | $2 / $6 | Proprietary |
| 8 | Mistral AI · EU | 88 | 80.5 | 80 | 80 | $0.70 / $2.10 | Open weights |
| 9 | xAI · US | 86 | 84.3 | 86 | 87 | $2 / $6 | Proprietary |
| 10 | Google · US | 84 | 89.7 | 92 | 86 | $2 / $10 | Proprietary |
| 11 | Moonshot AI · CN | 82 | 89.8 | 91 | 93 | $2 / $10 | Open weights |
| 12 | Anthropic · US | 70 | 93.3 | 98 | 99 | $4 / $20 | Proprietary |
| 13 | OpenAI · US | 62 | 90.1 | 95 | 91 | $10 / $50 | Proprietary |