Benchmark Almanac

ZIQI-Eval

GPT4 (gpt-4) 67.3% tops its leaderboard chatglm2-6b 26.1% leads the models we track
Not in model cards

Models answer multiple-choice questions about music history, performance, and theory. A second test asks them to choose the continuation of a melody.

The authors report 16 models. MuTheoryEval separately scored 15 configurations on a fixed sample of 500 comprehension questions.

Made by
Jiajia Li and collaborators
Released
February 2024
Kind
Exam questions
Level
University
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Subject
Arts
In and out
Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Qwen-14B #214 on the capability index
Last model added
September 2023
Scores refreshed
November 20th, 2026
17
Entries on its own leaderboard, 9 models we track

GPT4 (gpt-4) tops the leaderboard with 67.3% chatglm2-6b leads with 26.1% among models released by September 2023

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is chatglm2-6b, at 26.1%. This view shows only models we track, each at its best.

The newest model with a score here came out in September 2023, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT4 (gpt-4) Five-shot 67.3%
  2. 2 ChatGPT (gpt-3.5-turbo) Five-shot 45.6%
  3. 3 chatglm2-6b Five-shot 26.1%
  4. 4 ChatMusician-Base Zero-shot 24.6%
  5. 5 Ziya-LLaMA-13B-v1.1 Zero-shot 23.7%
  6. 6 Qwen-7B Zero-shot 20.4%
  7. 7 XVERSE-13B Five-shot 19.7%
  8. 8 Qwen-14B Zero-shot 18.9%
  9. 9 XVERSE-7B Zero-shot 15.8%
  10. 10 Baichuan 2-7B Zero-shot 11.8%
  11. 11 Baichuan 1-13B Zero-shot 11.5%
  12. 12 internlm-7b Zero-shot 8.22%
  13. 13 internlm-20b Five-shot 8.06%
  14. 14 Baichuan2-13B Five-shot 8.01%
  15. 15 Baichuan1-7B Zero-shot 6.90%
  16. 16 ChatMusician Five-shot 4.16%
  17. 17 educhat-base-002-13B Zero-shot 0%

Source: Official ZIQI-Eval leaderboard

#EntryScoreSettingDateSource
1 GPT4 (gpt-4) 67.3% Five-shot Official leaderboard
2 ChatGPT (gpt-3.5-turbo) 45.6% Five-shot Official leaderboard
3 chatglm2-6b 26.1% Five-shot Jun 2023 Official leaderboard
4 ChatMusician-Base 24.6% Zero-shot Official leaderboard
5 Ziya-LLaMA-13B-v1.1 23.7% Zero-shot Official leaderboard
6 Qwen-7B 20.4% Zero-shot Sep 2023 Official leaderboard
7 XVERSE-13B 19.7% Five-shot Official leaderboard
8 Qwen-14B 18.9% Zero-shot Sep 2023 Official leaderboard
9 XVERSE-7B 15.8% Zero-shot Official leaderboard
10 Baichuan 2-7B 11.8% Zero-shot Sep 2023 Official leaderboard
11 Baichuan 1-13B 11.5% Zero-shot Jul 2023 Official leaderboard
12 internlm-7b 8.22% Zero-shot Jul 2023 Official leaderboard
13 internlm-20b 8.06% Five-shot Sep 2023 Official leaderboard
14 Baichuan2-13B 8.01% Five-shot Sep 2023 Official leaderboard
15 Baichuan1-7B 6.90% Zero-shot Jun 2023 Official leaderboard
16 ChatMusician 4.16% Five-shot Official leaderboard
17 educhat-base-002-13B 0% Zero-shot Official leaderboard
#ModelScoreSettingDateSource
1 chatglm2-6b 26.1% Five-shot Jun 2023 Official leaderboard
2 Qwen-7B 20.4% Zero-shot Sep 2023 Official leaderboard
3 Qwen-14B 18.9% Zero-shot Sep 2023 Official leaderboard
4 Baichuan 2-7B 11.8% Zero-shot Sep 2023 Official leaderboard
5 Baichuan 1-13B 11.5% Zero-shot Jul 2023 Official leaderboard
6 internlm-7b 8.22% Zero-shot Jul 2023 Official leaderboard
7 internlm-20b 8.06% Five-shot Sep 2023 Official leaderboard
8 Baichuan2-13B 8.01% Five-shot Sep 2023 Official leaderboard
9 Baichuan1-7B 6.90% Zero-shot Jun 2023 Official leaderboard

Rankings by part

We read 10 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as 作曲理论, 世界民族音乐 and 流行音乐. chatglm2-6b leads 8 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

作曲理论

9 models we track have a score here.

  1. 1 Qwen-7B ↑1 19%
  2. 2 chatglm2-6b ↓1 18.1%
  3. 3 Qwen-14B 15.3%
  4. 4 Baichuan 1-13B ↑1 9.70%
  5. 5 Baichuan 2-7B ↓1 9.36%
  6. 6 internlm-7b 8.67%
  7. 7 Baichuan1-7B ↑2 7.64%
  8. 8 internlm-20b ↓1 6.78%
  9. 9 Baichuan2-13B ↓1 6.44%

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
chatglm2-6b
Zero-shot Five-shot
26.1%
Qwen-7B
Five-shot Zero-shot
20.4%
Qwen-14B
Five-shot
18.9%
Baichuan 2-7B
Five-shot Zero-shot
11.8%
Baichuan 1-13B
Five-shot Zero-shot
11.5%
internlm-7b
Five-shot
8.22%
internlm-20b
Zero-shot
8.06%
Baichuan2-13B
Zero-shot Five-shot
8.01%
Baichuan1-7B
Five-shot
6.90%

Source: Official ZIQI-Eval leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
chatglm2-6b Five-shot 26.1%
chatglm2-6b Zero-shot 24.1%
Qwen-7B Zero-shot 20.4%
Qwen-14B Zero-shot 18.9%
Qwen-14B Five-shot 17.7%
Baichuan 2-7B Zero-shot 11.8%
Baichuan 1-13B Zero-shot 11.5%
Baichuan 2-7B Five-shot 9.48%
internlm-7b Zero-shot 8.22%
internlm-20b Five-shot 8.06%
Baichuan2-13B Five-shot 8.01%
internlm-20b Zero-shot 7.43%
internlm-7b Five-shot 7.22%
Baichuan1-7B Zero-shot 6.90%
Baichuan 1-13B Five-shot 5.84%
Baichuan1-7B Five-shot 5.60%
Qwen-7B Five-shot 5.48%
Baichuan2-13B Zero-shot 5.23%

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
chatglm2-6b
Zero-shot Five-shot
26.1%
Qwen-7B
Five-shot Zero-shot
20.4%
Qwen-14B
Five-shot
18.9%
Baichuan 2-7B
Five-shot Zero-shot
11.8%
Baichuan 1-13B
Five-shot Zero-shot
11.5%
internlm-7b
Five-shot
8.22%
internlm-20b
Zero-shot
8.06%
Baichuan2-13B
Zero-shot Five-shot
8.01%
Baichuan1-7B
Five-shot
6.90%

Source: Official ZIQI-Eval leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
chatglm2-6b Five-shot 26.1%
chatglm2-6b Zero-shot 24.1%
Qwen-7B Zero-shot 20.4%
Qwen-14B Zero-shot 18.9%
Qwen-14B Five-shot 17.7%
Baichuan 2-7B Zero-shot 11.8%
Baichuan 1-13B Zero-shot 11.5%
Baichuan 2-7B Five-shot 9.48%
internlm-7b Zero-shot 8.22%
internlm-20b Five-shot 8.06%
Baichuan2-13B Five-shot 8.01%
internlm-20b Zero-shot 7.43%
internlm-7b Five-shot 7.22%
Baichuan1-7B Zero-shot 6.90%
Baichuan 1-13B Five-shot 5.84%
Baichuan1-7B Five-shot 5.60%
Qwen-7B Five-shot 5.48%
Baichuan2-13B Zero-shot 5.23%

Number one over time

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

Other labs
Own leaderboard Not tested since Feb 2024
chatglm2-6b 26.1%

Source: Official ZIQI-Eval leaderboard

Scores over time

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

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2023 Near the ceiling (85%) chatglm2-6b Baichuan1-7B 0% 20% 40% 60% 80% 100% 2023 Near the ceiling (85%) chatglm2-6b Baichuan1-7B

Source: Official ZIQI-Eval leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Knowledge

All 332 in Knowledge
  • MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects. Intern-S2-397B 89.8%
  • MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy. Seed1.8 92.3%
  • SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy. Qwen 3.6 Max (Preview) 73.9%
  • WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1. Tencent Hy4 preview 83.9%
  • BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly. Gemma 4 31B IT 74.4%
  • HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 3 years ago, in September 2023, for Qwen-7B. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (17 entries).
  • Pass Models scored 9 models we track, plus 8 other entries on its leaderboard.
  • Warning Newest score The newest came 3 years ago, in September 2023, for Qwen-7B.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Qwen-14B (#214).
  • Pass Headroom The best is 26.1%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 18 rows for models we track and 8 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check