DAXBenchDAX LLM Benchmarkv2.0
Updated Oct 18, 2026

Which LLM is best at DAX?

75 models tested on 94 DAX tasks: understanding, writing, and reasoning.

Developed by Maxim Anatsko in collaboration with SQLBI. Updated Oct 18, 2026.

GPT-6 Astra leads, solving 91 of 94 tasks.

  1. 1OpenAI logoOpenAI
    GPT-6 AstraAuto
    95.91%91 / 94 solved
  2. 2Anthropic logoAnthropic
    Claude Opus 5.5Auto
    95.38%90 / 94 solved
  3. 3OpenAI logoOpenAI
    GPT-5.6 SolMED
    91.46%87 / 94 solved

Model Leaderboard

Ranked by score

ModelTasks
90–100% solved
1
GPT-6 AstraAuto
OpenAI
95.91%
96.81%98.94%91/94
2
Claude Opus 5.5Auto
Anthropic
95.38%
95.74%97.87%90/94
3
GPT-5.6 SolMED
OpenAI
91.46%
92.55%96.81%87/94
4
GPT-6.1 SolAuto
OpenAI
91.42%
92.55%96.81%87/94
5
Grok 4.7Auto
SpaceXAI
91.22%
92.55%100%87/94
6
Claude Fable 5.1Auto
Anthropic
90.66%
91.49%97.87%86/94
80–89% solved
7
Gemini 4 ArgonAuto
Google
87.54%
88.3%98.94%83/94
8
GPT-5.5Auto
OpenAI
86.42%
88.3%95.74%83/94
9
Gemini 3.8 FlashAuto
Google
85.63%
87.23%100%82/94
10
Claude Opus 5Auto
Anthropic
85.4%
86.17%97.87%81/94
11
Claude Sonnet 5.5Auto
Anthropic
84.59%
86.17%97.87%81/94
12
Grok 4.6Auto
SpaceXAI
83.91%
86.17%93.62%81/94
13
GPT-6 LunaAuto
OpenAI
83%
85.11%93.62%80/94
14
Muse Spark 1.3Auto
Meta
82.57%
85.11%95.74%80/94
15
Muse WonderAuto
Meta
82.19%
84.04%92.55%79/94
16
Qwen3.8 Max (0902)Auto
Qwen
81.67%
82.98%97.87%78/94
17
Gemini 3.1 Pro PreviewAuto
Google
81.21%
82.98%96.81%78/94
18
MiMo-V2.6-ProAuto
Xiaomi1 looped
80.73%
81.91%91.49%77/94
19
Kimi K2.5Auto
Moonshot AI
79.74%
81.91%97.87%77/94
20
Claude Haiku 5.5Auto
Anthropic
79.37%
81.91%97.87%77/94
70–79% solved
21
Kimi K3Auto
Moonshot AI
78.05%
79.79%95.74%75/94
22
Kimi K2.7 CodeAuto
Moonshot AI
77.85%
79.79%98.94%75/94
23
GLM 5.3Auto
Z.AI1 truncated2 looped
77.29%
79.79%92.55%75/94
24
GLM 5.3 FlashAuto
Z.AI3 truncated3 looped
75.85%
78.72%91.49%74/94
25
GPT-5.4 MiniMED
OpenAI
75.76%
78.72%97.87%74/94
26
DeepSeek V4.1 Flash 1017Auto
DeepSeek5 looped
74.47%
77.66%92.55%73/94
27
DeepSeek V4.1 FlashAuto
DeepSeek1 truncated5 looped
72.83%
75.53%91.49%71/94
28
Ling 3.1 FlashAuto
inclusionAI4 looped
72.74%
74.47%90.43%70/94
29
Gemini 2.5 ProAuto
Google3 looped
71.94%
73.4%93.62%69/94
30
GLM 4.5Auto
Z.AI5 looped
71%
73.4%91.49%69/94
31
Claude Sonnet 5Auto
Anthropic
70.38%
72.34%96.81%68/94
32
GPT-5.1MED
OpenAI
70.37%
72.34%94.68%68/94
60–69% solved
33
DeepSeek V4 Pro 0813Auto
DeepSeek6 truncated
66.3%
69.15%86.17%65/94
34
GLM 5Auto
Z.AI1 truncated7 looped
66.03%
69.15%89.36%65/94
35
MiniMax M3Auto
Minimax
64.21%
67.02%90.43%63/94
36
Gemini 3 Flash PreviewAuto
Google
63.46%
67.02%93.62%63/94
37
Mistral Large 4Auto
Mistral1 truncated5 looped
63.16%
65.96%87.23%62/94
38
GPT-5 MiniAuto
OpenAI
62.96%
65.96%90.43%62/94
39
Muse Glimmer 30BAuto
Meta2 truncated
60.02%
63.83%90.43%60/94
40
Claude Opus 4.5Auto
Anthropic
59.59%
62.77%91.49%59/94
41
MiMo-V2.6-FlashAuto
Xiaomi
58.93%
62.77%93.62%59/94
42
Claude Sonnet 4Auto
Anthropic
58.42%
61.7%93.62%58/94
43
R1Auto
DeepSeek
57.65%
61.7%92.55%58/94
50–59% solved
44
Kimi K2 0711Auto
Moonshot AI
57.06%
59.57%88.3%56/94
45
Claude Sonnet 4.5Auto
Anthropic
55.32%
58.51%87.23%55/94
46
gpt-oss-120bAuto
OpenAI
54.96%
58.51%92.55%55/94
47
GPT-5.4 MiniAuto
OpenAI
54.41%
57.45%94.68%54/94
48
Gemma 4 26B A4B Auto
Google
53.25%
57.45%90.43%54/94
49
Mistral Large 2407Auto
Mistral
49.91%
54.26%90.43%51/94
50
Gemini 2.5 FlashAuto
Google
49.9%
54.26%91.49%51/94

About This Benchmark

Evaluation Method

Models are tested against DAX tasks of varying complexity using the Contoso sample dataset. Responses are evaluated for syntax correctness and output accuracy.

Scoring System

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.

Task Categories

Tasks cover aggregation, time intelligence, filtering, calculation, table manipulation, iterator, context transition across basic, intermediate, and advanced levels.

Browse by Category

Evaluating language model capabilities in DAX code generation

Developed by Maxim Anatsko in collaboration with SQLBI

maxanatsko