MLRC-Bench
Agents propose and implement methods for machine learning competitions. The scorer compares their performance with a supplied baseline and the strongest human solution.
- Made by
- Yunxiang Zhang and colleagues
- Released
- April 2025
- Kind
- AI research tasks
- Level
- Research frontier
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- Code sandbox
- Items from
- Exams and contests
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Agents
- Strongest model tested
- o3-mini #124 on the capability index
- Last model added
- March 2025
- Scores refreshed
- November 20th, 2026
gemini-exp-1206 tops the leaderboard with 9.30 Llama 3.1-405B leads with 6.26 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 on its own with several settings, we keep its best. Agents other companies built around a model get a pale shade of that model's lab color. Gray marks blends and entries not in our model list, such as fine-tunes and older models. None of these count toward leaders. The best model we track is Llama 3.1-405B, at 6.26. Agents built around a model, such as CoI-Agent (o1) + MLAB (gpt-4o) (GPT-4o + o1), are their own entries. We don't count them as the model's score. 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
- 1 gemini-exp-1206 9.30
- 2 CoI-Agent (o1) + MLAB (gpt-4o) (GPT-4o + o1) Mar 2025 7.11
- 3
Llama 3.1-405B 6.26
- 4 GPT-4o Mar 2025 5.41
- 5 o3-mini 4.20
- 6 Human Idea + MLAB (gpt-4o) + GPT-4o Mar 2025 3.53
- 7 Claude 3.5 Sonnet (October 2024) -5.19
Source: Official MLRC-Bench leaderboard
- 1
Llama 3.1-405B 6.26
- 2 GPT-4o Mar 2025 5.41
- 3 o3-mini 4.20
- 4 Claude 3.5 Sonnet (October 2024) -5.19
Source: Official MLRC-Bench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | gemini-exp-1206 | 9.30 | MLAB | Official leaderboard | |
| 2 | CoI-Agent (o1) + MLAB (gpt-4o) (GPT-4o + o1) Mar 2025 | 7.11 | Mar 2025 | Official leaderboard | |
| 3 | | 6.26 | MLAB | Jul 2024 | Official leaderboard |
| 4 | GPT-4o Mar 2025 | 5.41 | MLAB | Mar 2025 | Official leaderboard |
| 5 | o3-mini | 4.20 | MLAB | Jan 2025 | Official leaderboard |
| 6 | Human Idea + MLAB (gpt-4o) + GPT-4o Mar 2025 | 3.53 | Mar 2025 | Official leaderboard | |
| 7 | Claude 3.5 Sonnet (October 2024) | -5.19 | MLAB | Oct 2024 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 6.26 | MLAB | Jul 2024 | Official leaderboard |
| 2 | GPT-4o Mar 2025 | 5.41 | MLAB | Mar 2025 | Official leaderboard |
| 3 | o3-mini | 4.20 | MLAB | Jan 2025 | Official leaderboard |
| 4 | Claude 3.5 Sonnet (October 2024) | -5.19 | MLAB | Oct 2024 | Official leaderboard |
Rankings by part
We read 14 parts from the benchmark’s own leaderboard in 2 groups: relative improvement to human and absolute improvement to baseline. Each ranks models on one part of the test. Claude 3.5 Sonnet (October 2024) 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.
Temporal Action Localisation
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 0.80
- 2
Llama 3.1-405B ↓1 0.50
- 3 o3-mini 0.30
- 4 GPT-4o Mar 2025 ↓2 0.30
LLM Merging
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 5.00
- 2 GPT-4o Mar 2025 2.00
- 3
Llama 3.1-405B ↓2 -1.00
- 4 o3-mini ↓1 -1.00
Meta Learning
4 models we track have a score here.
- 1
Llama 3.1-405B -4.90
- 2 Claude 3.5 Sonnet (October 2024) ↑2 -4.90
- 3 o3-mini -4.90
- 4 GPT-4o Mar 2025 ↓2 -4.90
Next Product Recommendation
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 3.00
- 2 GPT-4o Mar 2025 0.60
- 3 o3-mini 0.10
- 4
Llama 3.1-405B ↓3 0.00
Rainfall Prediction
4 models we track have a score here.
- 1 GPT-4o Mar 2025 ↑1 47.5
- 2
Llama 3.1-405B ↓1 31.5
- 3 o3-mini 25.1
- 4 Claude 3.5 Sonnet (October 2024) 14.6
Machine Unlearning
4 models we track have a score here.
- 1
Llama 3.1-405B 6.20
- 2 o3-mini ↑1 3.60
- 3 GPT-4o Mar 2025 ↓1 -18.00
- 4 Claude 3.5 Sonnet (October 2024) -94.70
Backdoor Trigger Recovery
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 39.9
- 2
Llama 3.1-405B ↓1 11.5
- 3 GPT-4o Mar 2025 ↓1 10.4
- 4 o3-mini ↓1 6.20
Temporal Action Localisation
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 2.20
- 2
Llama 3.1-405B ↓1 1.50
- 3 o3-mini 0.90
- 4 GPT-4o Mar 2025 ↓2 0.90
LLM Merging
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 3.40
- 2 GPT-4o Mar 2025 1.40
- 3
Llama 3.1-405B ↓2 -0.70
- 4 o3-mini ↓1 -0.70
Meta Learning
4 models we track have a score here.
- 1
Llama 3.1-405B -14.90
- 2 Claude 3.5 Sonnet (October 2024) ↑2 -14.90
- 3 o3-mini -14.90
- 4 GPT-4o Mar 2025 ↓2 -14.90
Next Product Recommendation
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 12.3
- 2 GPT-4o Mar 2025 2.60
- 3 o3-mini 0.60
- 4
Llama 3.1-405B ↓3 0.00
Rainfall Prediction
4 models we track have a score here.
- 1 GPT-4o Mar 2025 ↑1 100.8
- 2
Llama 3.1-405B ↓1 66.7
- 3 o3-mini 53.3
- 4 Claude 3.5 Sonnet (October 2024) 31.0
Machine Unlearning
4 models we track have a score here.
- 1
Llama 3.1-405B 3.80
- 2 o3-mini ↑1 2.20
- 3 GPT-4o Mar 2025 ↓1 -11.10
- 4 Claude 3.5 Sonnet (October 2024) -58.60
Backdoor Trigger Recovery
4 models we track have a score here.
- 1 Claude 3.5 Sonnet (October 2024) ↑3 247.9
- 2
Llama 3.1-405B ↓1 71.7
- 3 GPT-4o Mar 2025 ↓1 64.5
- 4 o3-mini ↓1 38.8
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official MLRC-Bench 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 MLRC-Bench 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 AI research
- PaperBench In use Agent replicates ICML 2024 papers from scratch. LLM judge grades against author-written hierarchical rubrics.
Qwen 3.8 Max 93%
- MLS-Bench In use Agent must invent one algorithmic change in an ML research scaffold whose gain transfers across settings. Scored relative to human baselines. Claude Fable 5.1 50.3%
- PostTrainBench 1.2 In use CLI agent gets a base model, one H100 and a time budget to post-train it for a target benchmark. Scored on normalized gain. Claude Fable 5.1 44.6%
- MLE-bench In use Kaggle-style ML engineering competitions. Agent trains models and is scored against competition leaderboards. GPT-6.2 Astra 96.1%
- PMPP-Hard In use Agents write GPU kernels for 69 tasks that must be correct, and on 48 of them also run within 1.25 times the reference time. The score is the area under the solved-tasks curve over a 10-hour budget, out of 690 task-hours. Kimi K3 0.71
- KernelBench In use Replace PyTorch operators with faster custom GPU kernels. Scored on correctness and speedup.
MiniMax-M3 28.8%
Data audit
8 of 9 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 2 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (7 entries).
- Pass Models scored 4 models we track, plus 3 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 o3-mini (#124).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on huggingface.co: 6 rows for models we track and 1 other entry 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.