Thorbase
[Power your AI workloads]
Plug-and-play model aggregation with enterprise-grade controls.
Models
Explore our current model offerings.
GLM-5
Zhipu
In
TextImageOut
TextInput$0.60 / 1M
Output$2.10 / 1M
Context202,800
Parameters355B
DeepSeek V3.1
DeepSeek
In
TextOut
TextInput$0.20 / 1M
Output$0.80 / 1M
Context131,072
Parameters671B MoE
DeepSeek V3.2
DeepSeek
In
TextOut
TextInput$0.20 / 1M
Output$0.30 / 1M
Context131,072
Parameters671B MoE
Qwen3.5-397B-A17B
Alibaba
In
TextImageOut
TextInput$0.40 / 1M
Output$2.50 / 1M
Context131,072
Parameters72B
Kimi K2.5
Moonshot
In
TextFileOut
TextInput$0.40 / 1M
Output$2.10 / 1M
Context131,072
Parameters1T MoE
Model ranking leaderboard
Benchmark scores from artificialanalysis.ai
| Rank | Model | Selected benchmark | Avg price / 1M | Value index |
|---|---|---|---|---|
| 1 | GLM-5 | 91/100 | $0.62 | 146.8 |
| 2 | Llama 3.3 70B | 79/100 | $0.55 | 143.6 |
| 3 | DeepSeek V3 | 86/100 | $0.74 | 116.2 |
| 4 | Qwen 2.5 72B | 84/100 | $1.10 | 76.4 |
| 5 | Kimi K2 | 82/100 | $1.80 | 45.6 |
| 6 | MiniMax Text 01 | 76/100 | $2.60 | 29.2 |
| 7 | Gemini 2.5 Pro | 95/100 | $6.20 | 15.3 |
| 8 | GPT-5 | 90/100 | $7.50 | 12.0 |
| 9 | Claude Sonnet 4 | 93/100 | $9.00 | 10.3 |
Value index = benchmark score รท average price per 1M tokens (higher means more score per dollar). Bars still reflect raw benchmark only; use the index to compare cost-adjusted appeal.