Benchmark Almanac

DAXBench v2

GPT-6 Astra 95.9% leads since October 2026
Not in model cards

Models write DAX, the formula language of Power BI, for 94 tasks on understanding, writing and reasoning over the Contoso sample model. Each answer runs inside real DAX queries in several report contexts, such as detail rows, subtotals, the grand total and slicers. Harder tasks earn more points, with bonuses for best practices and efficient code.

Version 1 (December 2025) had 30 tasks, so its scores do not compare with v2. Models answer in one shot without running their code. The board listed 80 models on November 4th, 2026. SQLBI announced v2 on its blog on October 27th, 2026.

Released
October 2026
Kind
Data science and SQL
Level
University
Scored
Checked automatically
Answers
Code
Strongest model tested
Claude Fable 5.2 #1 on the capability index
Last model added
November 2026
Scores refreshed
November 12th, 2026
50
Entries on its own leaderboard, 47 models we track

GPT-6 Astra leads with 95.9%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

50 entries. Scroll sideways for the rest.

  1. 95.9 GPT-6 Astra
  2. 95.4 Claude Opus 5.5
  3. 91.9 GPT-6.1 Terra
  4. 91.5 GPT-5.6 Sol
  5. 91.4 GPT-6.1 Sol
  6. 91.2 Grok 4.7
  7. 90.7 Claude Fable 5.1
  8. 87.8 Claude Fable 5.2
  9. 87.5 Gemini 4 Argon
  10. 86.4 GPT-5.5
  11. 85.7 Tencent Hy4
  12. 85.6 Gemini 3.8 Flash
  13. 85.4 Claude Opus 5
  14. 84.6 Claude Sonnet 5.5
  15. 83.9 Grok 4.6
  16. 83.9 Kimi K3.1 Pro
  17. 83.0 GPT-6 Luna
  18. 82.6 Muse Spark 1.3
  19. 82.2 Muse Wonder
  20. 81.7 Qwen3.8 Max (0902)
  21. 81.2 Gemini 3.1 Pro
  22. 80.7 MiMo-V2.6 Pro
  23. 79.7 Kimi K2.5
  24. 79.4 Claude Haiku 5.5
  25. 78.0 Kimi K3
  26. 77.8 Kimi K2.7 Code
  27. 77.3 GLM-5.3
  28. 75.8 GLM-5.3-Flash
  29. 75.8 GPT-5.4 Mini
  30. 74.5 DeepSeek V4.1 Flash 1017
  31. 72.8 DeepSeek V4.1 Flash
  32. 72.7 Ling-3.1-flash
  33. 71.9 Gemini 2.5 Pro (snapshot unspecified)
  34. 71.0 GLM-4.5
  35. 70.4 Claude Sonnet 5
  36. 70.4 GPT-5.1
  37. 66.3 DeepSeek V4 Pro 0813
  38. 66.0 GLM-5
  39. 64.2 MiniMax-M3
  40. 63.5 Gemini 3 Flash
  41. 63.2 Mistral Large 4
  42. 63.0 GPT-5 mini
  43. 61.0 MiniMax-M3.1-Flash
  44. 60.0 Muse Glimmer 30B
  45. 59.6 Claude Opus 4.5
  46. 59.4 Muse Ember
  47. 59.0 Qwen3.6 27B
  48. 58.9 MiMo-V2.6 Flash
  49. 58.4 Claude Sonnet 4
  50. 57.6 DeepSeek-R1 (snapshot unspecified)
  1. 1 GPT-6 Astra Auto 95.9%
  2. 2 Claude Opus 5.5 Auto 95.4%
  3. 3 GPT-6.1 Terra Auto 91.9%
  4. 4 GPT-5.6 Sol MED 91.5%
  5. 5 GPT-6.1 Sol Auto 91.4%
  6. 6 Grok 4.7 Auto 91.2%
  7. 7 Claude Fable 5.1 Auto 90.7%
  8. 8 Claude Fable 5.2 Auto 87.8%
  9. 9 Gemini 4 Argon Auto 87.5%
  10. 10 GPT-5.5 Auto 86.4%
  11. 11 Tencent Hy4 Auto 85.7%
  12. 12 Gemini 3.8 Flash Auto 85.6%
  13. 13 Claude Opus 5 Auto 85.4%
  14. 14 Claude Sonnet 5.5 Auto 84.6%
  15. 15 Grok 4.6 Auto 83.9%
  16. 16 Kimi K3.1 Pro Auto 83.9%
  17. 17 GPT-6 Luna Auto 83%
  18. 18 Muse Spark 1.3 Auto 82.6%
  19. 19 Muse Wonder Auto 82.2%
  20. 20 Qwen3.8 Max (0902) Auto 81.7%
  21. 21 Gemini 3.1 Pro Auto 81.2%
  22. 22 MiMo-V2.6 Pro Auto 80.7%
  23. 23 Kimi K2.5 Auto 79.7%
  24. 24 Claude Haiku 5.5 Auto 79.4%
  25. 25 Kimi K3 Auto 78%
  26. 26 Kimi K2.7 Code Auto 77.8%
  27. 27 GLM-5.3 Auto 77.3%
  28. 28 GLM-5.3-Flash Auto 75.8%
  29. 29 GPT-5.4 Mini MED 75.8%
  30. 30 DeepSeek V4.1 Flash 1017 Auto 74.5%
  31. 31 DeepSeek V4.1 Flash Auto 72.8%
  32. 32 Ling-3.1-flash Auto 72.7%
  33. 33 Gemini 2.5 Pro (snapshot unspecified) Auto 71.9%
  34. 34 GLM-4.5 Auto 71%
  35. 35 Claude Sonnet 5 Auto 70.4%
  36. 36 GPT-5.1 MED 70.4%
  37. 37 DeepSeek V4 Pro 0813 Auto 66.3%
  38. 38 GLM-5 Auto 66%
  39. 39 MiniMax-M3 Auto 64.2%
  40. 40 Gemini 3 Flash Auto 63.5%
  41. 41 Mistral Large 4 Auto 63.2%
  42. 42 GPT-5 mini Auto 63%
  43. 43 MiniMax-M3.1-Flash Auto 61%
  44. 44 Muse Glimmer 30B Auto 60%
  45. 45 Claude Opus 4.5 Auto 59.6%
  46. 46 Muse Ember Auto 59.4%
  47. 47 Qwen3.6 27B Auto 59%
  48. 48 MiMo-V2.6 Flash Auto 58.9%
  49. 49 Claude Sonnet 4 Auto 58.4%
  50. 50 DeepSeek-R1 (snapshot unspecified) Auto 57.6%

Source: Official DAXBench v2 leaderboard

#EntryScoreSettingDateSource
1 GPT-6 Astra 95.9% Auto Sep 2026 Official leaderboard
2 Claude Opus 5.5 95.4% Auto Sep 2026 Official leaderboard
3 GPT-6.1 Terra 91.9% Auto Oct 2026 Official leaderboard
4 GPT-5.6 Sol 91.5% MED Jul 2026 Official leaderboard
5 GPT-6.1 Sol 91.4% Auto Sep 2026 Official leaderboard
6 Grok 4.7 91.2% Auto Sep 2026 Official leaderboard
7 Claude Fable 5.1 90.7% Auto Sep 2026 Official leaderboard
8 Claude Fable 5.2 87.8% Auto Nov 2026 Official leaderboard
9 Gemini 4 Argon 87.5% Auto Sep 2026 Official leaderboard
10 GPT-5.5 86.4% Auto Apr 2026 Official leaderboard
11 Tencent Hy4 85.7% Auto Nov 2026 Official leaderboard
12 Gemini 3.8 Flash 85.6% Auto Sep 2026 Official leaderboard
13 Claude Opus 5 85.4% Auto Jul 2026 Official leaderboard
14 Claude Sonnet 5.5 84.6% Auto Sep 2026 Official leaderboard
15 Grok 4.6 83.9% Auto Aug 2026 Official leaderboard
16 Kimi K3.1 Pro 83.9% Auto Oct 2026 Official leaderboard
17 GPT-6 Luna 83% Auto Sep 2026 Official leaderboard
18 Muse Spark 1.3 82.6% Auto Sep 2026 Official leaderboard
19 Muse Wonder 82.2% Auto Oct 2026 Official leaderboard
20 Qwen3.8 Max (0902) 81.7% Auto Sep 2026 Official leaderboard
21 Gemini 3.1 Pro 81.2% Auto Feb 2026 Official leaderboard
22 MiMo-V2.6 Pro 80.7% Auto Sep 2026 Official leaderboard
23 Kimi K2.5 79.7% Auto Jan 2026 Official leaderboard
24 Claude Haiku 5.5 79.4% Auto Oct 2026 Official leaderboard
25 Kimi K3 78% Auto Jul 2026 Official leaderboard
26 Kimi K2.7 Code 77.8% Auto Jun 2026 Official leaderboard
27 GLM-5.3 77.3% Auto Aug 2026 Official leaderboard
28 GLM-5.3-Flash 75.8% Auto Aug 2026 Official leaderboard
29 GPT-5.4 Mini 75.8% MED Mar 2026 Official leaderboard
30 DeepSeek V4.1 Flash 1017 74.5% Auto Oct 2026 Official leaderboard
31 DeepSeek V4.1 Flash 72.8% Auto Sep 2026 Official leaderboard
32 Ling-3.1-flash 72.7% Auto Oct 2026 Official leaderboard
33 Gemini 2.5 Pro (snapshot unspecified) 71.9% Auto Official leaderboard
34 GLM-4.5 71% Auto Aug 2025 Official leaderboard
35 Claude Sonnet 5 70.4% Auto Jun 2026 Official leaderboard
36 GPT-5.1 70.4% MED Nov 2025 Official leaderboard
37 DeepSeek V4 Pro 0813 66.3% Auto Aug 2026 Official leaderboard
38 GLM-5 66% Auto Feb 2026 Official leaderboard
39 MiniMax-M3 64.2% Auto Jun 2026 Official leaderboard
40 Gemini 3 Flash 63.5% Auto Dec 2025 Official leaderboard
41 Mistral Large 4 63.2% Auto Oct 2026 Official leaderboard
42 GPT-5 mini 63% Auto Aug 2025 Official leaderboard
43 MiniMax-M3.1-Flash 61% Auto Oct 2026 Official leaderboard
44 Muse Glimmer 30B 60% Auto Official leaderboard
45 Claude Opus 4.5 59.6% Auto Nov 2025 Official leaderboard
46 Muse Ember 59.4% Auto Nov 2026 Official leaderboard
47 Qwen3.6 27B 59% Auto Apr 2026 Official leaderboard
48 MiMo-V2.6 Flash 58.9% Auto Sep 2026 Official leaderboard
49 Claude Sonnet 4 58.4% Auto May 2025 Official leaderboard
50 DeepSeek-R1 (snapshot unspecified) 57.6% Auto Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard Passed 85% in Oct 2026, by GPT-6 Astra
GPT-6 Astra 95.9%

Source: Official DAXBench v2 leaderboard

GPT-5.5 on the official leaderboard passed 85% in April 2026

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) GPT-6 Astra GPT-5.6 Sol GPT-5.5 Gemini 3.1 Pro GLM-4.5 Claude Sonnet 4 0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) GPT-6 Astra GPT-5.5

Source: Official DAXBench v2 leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Coding

All 282 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Fable 5.2 96.6%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471