ray-settlement-queries · Models
Which model should Dvar use to score your traffic?
The judge, topic names, rubric drafts and alternative replies call these models with your tested key, through the Dvar connector.
Models on Dvar Inference
Open-weight models served from India. Tested ones show their share of benchmark chats resolved; the judge that scores the two subjective checks is google/gemini-4-argon.
| Model | Price per 1M tokens (in / out) | On this benchmark |
|---|---|---|
| deepseek-v4-flash-0731 dvar:deepseek-v4-flash-0731 | ₹10 / ₹20 | 78.8% fully right |
| gemma-4-26b-a4b-it dvar:gemma-4-26b-a4b-it | ₹10 / ₹30 | not tested |
| gemma-4-31b-it dvar:gemma-4-31b-it | ₹10 / ₹30 | not tested |
| glm-5.3 dvar:glm-5.3 | ₹105 / ₹340 | not tested |
| glm-5.3-flash dvar:glm-5.3-flash | ₹10 / ₹40 | 79.4% fully right |
| gpt-oss-120b dvar:gpt-oss-120b | ₹10 / ₹40 | 58.8% fully right |
| muse-glimmer-30b dvar:muse-glimmer-30b | ₹25 / ₹110 | wrote the synthetic messages |
| nemotron-3.5-lightning-30b-a3b-bf16 dvar:nemotron-3.5-lightning-30b-a3b-bf16 | ₹5 / ₹20 | 43.1% fully right |
| qwen-3.8-27b-fp8 dvar:qwen-3.8-27b-fp8 | ₹25 / ₹210 | not tested |