CIBench
Agents use code interpreters to analyze data. The evaluation checks stages of the interactive analysis.
- Made by
- OpenCompass and collaborators
- Released
- July 2024
- Kind
- Data science and SQL
- Level
- University
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Multi-agent
- Environment
- Code sandbox
- Items from
- Generated, Written by experts
- 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
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:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 75.5 GPT-4 Turbo
- 74.5 GPT-4o
- 65.5 Llama-3-70B-Instruct
- 65.1 Qwen-72B-Chat
- 64.4 Llama-3-8B-Instruct
- 60.8 DeepSeek-67B-Chat
- 60.6 InternLM2-20B-Chat
- 60.4 Mixtral-8x7B-Instruct
- 59.8
Qwen1.5-14B-Chat
- 57.5
Qwen-14B
- 56.0
DeepSeek Coder 6.7B
- 53.0 Yi-34B-Chat
- 50.7 InternLM2-7B-Chat
- 49.3
Qwen1.5-7B-Chat
- 48.3
Mistral-7B-Instruct-v0.1
- 48.1 Qwen-7B-Chat
- 39.6 Vicuna-13B-v1.5-16k
- 33.7 DeepSeek-7B-Chat
- 32.7 Vicuna-7B-v1.5-16k
- 32.6 ChatGLM3-6B-32K
- 31.0
Yi-6B-Chat
- 29.9 Llama-2-70B-Chat
- 23.4 Llama-2-13B-Chat
- 19.4 Llama-2-7B-Chat
- 1 GPT-4 Turbo Nov 2023 75.5%
- 2 GPT-4o Mar 2025 74.5%
- 3 Llama-3-70B-Instruct 65.5%
- 4 Qwen-72B-Chat 65.1%
- 5 Llama-3-8B-Instruct 64.4%
- 6 DeepSeek-67B-Chat 60.8%
- 7 InternLM2-20B-Chat 60.6%
- 8 Mixtral-8x7B-Instruct 60.4%
- 9
Qwen1.5-14B-Chat 59.8%
- 10
Qwen-14B 57.5%
- 11
DeepSeek Coder 6.7B 56%
- 12 Yi-34B-Chat 53%
- 13 InternLM2-7B-Chat 50.7%
- 14
Qwen1.5-7B-Chat 49.3%
- 15
Mistral-7B-Instruct-v0.1 48.3%
- 16 Qwen-7B-Chat 48.1%
- 17 Vicuna-13B-v1.5-16k 39.6%
- 18 DeepSeek-7B-Chat 33.7%
- 19 Vicuna-7B-v1.5-16k 32.7%
- 20 ChatGLM3-6B-32K 32.6%
- 21
Yi-6B-Chat 31%
- 22 Llama-2-70B-Chat 29.9%
- 23 Llama-2-13B-Chat 23.4%
- 24 Llama-2-7B-Chat 19.4%
Source: Official CIBench leaderboard
- 1 GPT-4 Turbo Nov 2023 75.5%
- 2 GPT-4o Mar 2025 74.5%
- 3
Qwen1.5-14B-Chat 59.8%
- 4
Qwen-14B 57.5%
- 5
DeepSeek Coder 6.7B 56%
- 6
Qwen1.5-7B-Chat 49.3%
- 7
Mistral-7B-Instruct-v0.1 48.3%
- 8
Yi-6B-Chat 31%
Source: Official CIBench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 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 | | 59.8% | Feb 2024 | Official leaderboard | |
| 10 | | 57.5% | Sep 2023 | Official leaderboard | |
| 11 | | 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 | | 49.3% | Feb 2024 | Official leaderboard | |
| 15 | | 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 | | 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 |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-4 Turbo Nov 2023 | 75.5% | Nov 2023 | Official leaderboard | |
| 2 | GPT-4o Mar 2025 | 74.5% | Mar 2025 | Official leaderboard | |
| 3 | | 59.8% | Feb 2024 | Official leaderboard | |
| 4 | | 57.5% | Sep 2023 | Official leaderboard | |
| 5 | | 56% | Nov 2023 | Official leaderboard | |
| 6 | | 49.3% | Feb 2024 | Official leaderboard | |
| 7 | | 48.3% | Sep 2023 | Official leaderboard | |
| 8 | | 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 GPT-4o Mar 2025 ↑1 100%
- 2 GPT-4 Turbo Nov 2023 ↓1 99.8%
- 3
Qwen1.5-14B-Chat 99.7%
- 4
Qwen1.5-7B-Chat ↑2 99.1%
- 5
Mistral-7B-Instruct-v0.1 ↑2 97.5%
- 6
Qwen-14B ↓2 93.1%
- 7
Yi-6B-Chat ↑1 90.6%
- 8
DeepSeek Coder 6.7B ↓3 75.9%
E-Exe
8 models we track have a score here.
- 1 GPT-4 Turbo Nov 2023 98.6%
- 2 GPT-4o Mar 2025 97.2%
- 3
Qwen1.5-14B-Chat 84.8%
- 4
Qwen-14B 70.9%
- 5
Qwen1.5-7B-Chat ↑1 69.9%
- 6
Mistral-7B-Instruct-v0.1 ↑1 67.2%
- 7
DeepSeek Coder 6.7B ↓2 66%
- 8
Yi-6B-Chat 50.2%
E-Num
8 models we track have a score here.
- 1 GPT-4 Turbo Nov 2023 77.8%
- 2 GPT-4o Mar 2025 74.1%
- 3
Qwen1.5-14B-Chat 56.9%
- 4
Qwen-14B 48.2%
- 5
Mistral-7B-Instruct-v0.1 ↑2 41.9%
- 6
DeepSeek Coder 6.7B ↓1 40.7%
- 7
Qwen1.5-7B-Chat ↓1 35.4%
- 8
Yi-6B-Chat 22.5%
E-Text
8 models we track have a score here.
- 1 GPT-4o Mar 2025 ↑1 79%
- 2 GPT-4 Turbo Nov 2023 ↓1 78.9%
- 3
Qwen1.5-14B-Chat 63.3%
- 4
Qwen-14B 60.7%
- 5
Qwen1.5-7B-Chat ↑1 59.1%
- 6
Mistral-7B-Instruct-v0.1 ↑1 53.6%
- 7
DeepSeek Coder 6.7B ↓2 48.5%
- 8
Yi-6B-Chat 21.7%
E-Vis
8 models we track have a score here.
- 1 GPT-4 Turbo Nov 2023 64%
- 2 GPT-4o Mar 2025 61.1%
- 3
Qwen1.5-14B-Chat 51.4%
- 4
Qwen-14B 48.7%
- 5
DeepSeek Coder 6.7B 47.7%
- 6
Mistral-7B-Instruct-v0.1 ↑1 41.7%
- 7
Qwen1.5-7B-Chat ↓1 39.1%
- 8
Yi-6B-Chat 31.1%
O-Tool
8 models we track have a score here.
- 1
Qwen1.5-14B-Chat ↑2 99.9%
- 2
Qwen1.5-7B-Chat ↑4 99.8%
- 3
Yi-6B-Chat ↑5 99.7%
- 4 GPT-4 Turbo Nov 2023 ↓3 99.6%
- 5
Qwen-14B ↓1 99.2%
- 6
Mistral-7B-Instruct-v0.1 ↑1 99.2%
- 7 GPT-4o Mar 2025 ↓5 99.2%
- 8
DeepSeek Coder 6.7B ↓3 96.8%
O-Exe
8 models we track have a score here.
- 1 GPT-4 Turbo Nov 2023 97.7%
- 2 GPT-4o Mar 2025 94.8%
- 3
Qwen1.5-14B-Chat 92.7%
- 4
DeepSeek Coder 6.7B ↑1 83%
- 5
Qwen-14B ↓1 82.4%
- 6
Qwen1.5-7B-Chat 75.9%
- 7
Mistral-7B-Instruct-v0.1 75.7%
- 8
Yi-6B-Chat 61.5%
O-Num
8 models we track have a score here.
- 1 GPT-4o Mar 2025 ↑1 81.7%
- 2 GPT-4 Turbo Nov 2023 ↓1 77.5%
- 3
Qwen1.5-14B-Chat 61.7%
- 4
DeepSeek Coder 6.7B ↑1 59.2%
- 5
Qwen-14B ↓1 55.5%
- 6
Qwen1.5-7B-Chat 48.6%
- 7
Mistral-7B-Instruct-v0.1 48%
- 8
Yi-6B-Chat 29.5%
O-Text
8 models we track have a score here.
- 1 GPT-4 Turbo Nov 2023 81.6%
- 2 GPT-4o Mar 2025 79.7%
- 3
DeepSeek Coder 6.7B ↑2 77.2%
- 4
Qwen-14B 74.6%
- 5
Qwen1.5-14B-Chat ↓2 62.6%
- 6
Qwen1.5-7B-Chat 61.7%
- 7
Mistral-7B-Instruct-v0.1 55.4%
- 8
Yi-6B-Chat 40.8%
O-Vis
8 models we track have a score here.
- 1 GPT-4 Turbo Nov 2023 73.1%
- 2 GPT-4o Mar 2025 71.6%
- 3
Qwen1.5-14B-Chat 63.1%
- 4
DeepSeek Coder 6.7B ↑1 62.5%
- 5
Qwen-14B ↓1 57.3%
- 6
Qwen1.5-7B-Chat 52.1%
- 7
Mistral-7B-Instruct-v0.1 49.1%
- 8
Yi-6B-Chat 40.2%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
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.
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
- 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.