Benchmark Almanac

CIBench

GPT-4 Turbo Nov 2023 75.5% leads since July 2024
Not in model cards

Agents use code interpreters to analyze data. The evaluation checks stages of the interactive analysis.

Released
July 2024
Kind
Data science and SQL
Level
University
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Tables, Code → Actions, Code, Text
Interaction
Multi-agent
Environment
Code sandbox
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
GPT-4 Turbo Nov 2023 #176 on the capability index
Last model added
March 2025
Scores refreshed
November 20th, 2026
24
Entries on its own leaderboard, 8 models we track

GPT-4 Turbo leads with 75.5% among models released by March 2025 GPT-4 Turbo leads with 75.5% among models released by March 2025

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. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in March 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-4 Turbo Nov 2023 75.5%
  2. 2 GPT-4o Mar 2025 74.5%
  3. 3 Llama-3-70B-Instruct 65.5%
  4. 4 Qwen-72B-Chat 65.1%
  5. 5 Llama-3-8B-Instruct 64.4%
  6. 6 DeepSeek-67B-Chat 60.8%
  7. 7 InternLM2-20B-Chat 60.6%
  8. 8 Mixtral-8x7B-Instruct 60.4%
  9. 9 Qwen1.5-14B-Chat 59.8%
  10. 10 Qwen-14B 57.5%
  11. 11 DeepSeek Coder 6.7B 56%
  12. 12 Yi-34B-Chat 53%
  13. 13 InternLM2-7B-Chat 50.7%
  14. 14 Qwen1.5-7B-Chat 49.3%
  15. 15 Mistral-7B-Instruct-v0.1 48.3%
  16. 16 Qwen-7B-Chat 48.1%
  17. 17 Vicuna-13B-v1.5-16k 39.6%
  18. 18 DeepSeek-7B-Chat 33.7%
  19. 19 Vicuna-7B-v1.5-16k 32.7%
  20. 20 ChatGLM3-6B-32K 32.6%
  21. 21 Yi-6B-Chat 31%
  22. 22 Llama-2-70B-Chat 29.9%
  23. 23 Llama-2-13B-Chat 23.4%
  24. 24 Llama-2-7B-Chat 19.4%

Source: Official CIBench leaderboard

#EntryScoreSettingDateSource
1 GPT-4 Turbo Nov 2023 75.5% Nov 2023 Official leaderboard
2 GPT-4o Mar 2025 74.5% Mar 2025 Official leaderboard
3 Llama-3-70B-Instruct open weights 65.5% Official leaderboard
4 Qwen-72B-Chat open weights 65.1% Official leaderboard
5 Llama-3-8B-Instruct open weights 64.4% Official leaderboard
6 DeepSeek-67B-Chat open weights 60.8% Official leaderboard
7 InternLM2-20B-Chat open weights 60.6% Official leaderboard
8 Mixtral-8x7B-Instruct open weights 60.4% Official leaderboard
9 Qwen1.5-14B-Chat 59.8% Feb 2024 Official leaderboard
10 Qwen-14B 57.5% Sep 2023 Official leaderboard
11 DeepSeek Coder 6.7B 56% Nov 2023 Official leaderboard
12 Yi-34B-Chat open weights 53% Official leaderboard
13 InternLM2-7B-Chat open weights 50.7% Official leaderboard
14 Qwen1.5-7B-Chat 49.3% Feb 2024 Official leaderboard
15 Mistral-7B-Instruct-v0.1 48.3% Sep 2023 Official leaderboard
16 Qwen-7B-Chat open weights 48.1% Official leaderboard
17 Vicuna-13B-v1.5-16k open weights 39.6% Official leaderboard
18 DeepSeek-7B-Chat open weights 33.7% Official leaderboard
19 Vicuna-7B-v1.5-16k open weights 32.7% Official leaderboard
20 ChatGLM3-6B-32K open weights 32.6% Official leaderboard
21 Yi-6B-Chat 31% Nov 2023 Official leaderboard
22 Llama-2-70B-Chat open weights 29.9% Official leaderboard
23 Llama-2-13B-Chat open weights 23.4% Official leaderboard
24 Llama-2-7B-Chat open weights 19.4% Official leaderboard
#ModelScoreSettingDateSource
1 GPT-4 Turbo Nov 2023 75.5% Nov 2023 Official leaderboard
2 GPT-4o Mar 2025 74.5% Mar 2025 Official leaderboard
3 Qwen1.5-14B-Chat 59.8% Feb 2024 Official leaderboard
4 Qwen-14B 57.5% Sep 2023 Official leaderboard
5 DeepSeek Coder 6.7B 56% Nov 2023 Official leaderboard
6 Qwen1.5-7B-Chat 49.3% Feb 2024 Official leaderboard
7 Mistral-7B-Instruct-v0.1 48.3% Sep 2023 Official leaderboard
8 Yi-6B-Chat 31% Nov 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 E-Tool, E-Exe and E-Num. GPT-4 Turbo leads 6 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.

E-Tool

8 models we track have a score here.

  1. 1 GPT-4o Mar 2025 ↑1 100%
  2. 2 GPT-4 Turbo Nov 2023 ↓1 99.8%
  3. 3 Qwen1.5-14B-Chat 99.7%
  4. 4 Qwen1.5-7B-Chat ↑2 99.1%
  5. 5 Mistral-7B-Instruct-v0.1 ↑2 97.5%
  6. 6 Qwen-14B ↓2 93.1%
  7. 7 Yi-6B-Chat ↑1 90.6%
  8. 8 DeepSeek Coder 6.7B ↓3 75.9%

Number one over time

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

OpenAI
Own leaderboard Not tested since Mar 2025
GPT-4 Turbo 75.5%

Source: Official CIBench 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% 2024 2025 Near the ceiling (85%) GPT-4 Turbo Qwen-14B 0% 20% 40% 60% 80% 100% 2024 2025 Near the ceiling (85%) GPT-4 Turbo Qwen-14B

Source: Official CIBench 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 Coding

All 523 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

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 20 months ago, in March 2025, for GPT-4o. 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 (24 entries).
  • Pass Models scored 8 models we track, plus 16 other entries on its leaderboard.
  • Warning Newest score The newest came 20 months ago, in March 2025, for GPT-4o.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-4 Turbo (#176).
  • Pass Headroom The best is 75.5%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on open-compass.github.io: 8 rows for models we track and 16 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