Which LLM is best at DAX?
DAXBench tests how models understand, write, and reason about DAX.
Methodology designed by Maxim Anatsko.
Last updated: Oct 12, 2026
11 models · 94 tasks · DAX Benchmark v2.0
Ranked by score
| Model | |||||
|---|---|---|---|---|---|
90–100% solved | |||||
1 | Claude Opus 5.5Auto Anthropic | 95.38% | |||
2 | GPT-6.1 SolAuto OpenAI | 91.42% | |||
3 | Grok 4.7Auto SpaceXAI | 91.22% | |||
80–89% solved | |||||
4 | Gemini 3.8 FlashAuto Google | 85.63% | |||
5 | Claude Sonnet 5.5Auto Anthropic | 84.59% | |||
6 | GPT-6 LunaAuto OpenAI | 83% | |||
7 | Gemini 3.1 Pro PreviewAuto Google | 81.21% | |||
8 | Claude Haiku 5.5Auto Anthropic | 79.37% | |||
70–79% solved | |||||
9 | GLM 5.3 FlashAuto Z.AI3 truncated3 looped | 75.85% | |||
10 | DeepSeek V4.1 FlashAuto DeepSeek1 truncated5 looped | 72.83% | |||
60–69% solved | |||||
11 | DeepSeek V4 Pro 0813Auto DeepSeek6 truncated | 66.3% | |||
Models are tested against DAX tasks of varying complexity using the Contoso sample dataset. Responses are evaluated for syntax correctness and output accuracy.
Harder tasks are worth more points. Correct solutions also earn bonus points for following DAX best practices, writing efficient code, and producing clear, readable output.
Tasks cover aggregation, time intelligence, filtering, calculation, table manipulation, iterator, context transition across basic, intermediate, and advanced levels.