USACO
The USACO benchmark evaluates AI agents on competitive programming problems from the USA Computing Olympiad. It consists of 307 problems, complete with exhaustive test cases, problem analyses, and difficulty labels.
Paper: Can Language Models Solve Olympiad Programming? (Shi et al., 2024)
Key Features of USACO
Difficulty Levels
Tasks span Bronze to Platinum difficulty levels, requiring knowledge of data structures and algorithms.
Real Competition Tasks
Problems are sourced from actual USACO competitions, consisting of problems challenging for human programmers.
USACO Leaderboard
Rank | Agent | Models |
Verified
Verified Results
Results have been reproduced by the HAL team |
Accuracy
Accuracy
Confidence intervals show the min-max values across runs for those agents where multiple runs are available |
Cost (USD)
Total Cost
Total API cost for running the agent on all tasks. Confidence intervals show the min-max values across runs for those agents where multiple runs are available |
Runs
Number of Runs
The number of runs for this agent submitted to the leaderboard. To submit multiple evaluations, rerun the same agent and set the same agent name |
Traces |
---|---|---|---|---|---|---|---|
1 | o3-mini-2025-01-31 med. | ✓ | 38.76% | $19.68 | 1 | Download | |
2 | o3-mini-2025-01-31 med. | ✓ | 36.16% | $9.58 | 1 | Download | |
3 | claude-3-7-sonnet-20250219 | ✓ | 28.99% | $38.94 | 1 | Download | |
4 | gpt-4.5-preview-2025-02-27 | ✓ | 27.04% | $367.27 | 1 | Download | |
5 | o1-mini-2024-09-12 | ✓ | 23.78% | $13.76 | 1 | Download | |
6 | claude-3-7-sonnet-20250219 | ✓ | 19.54% | $8.77 | 1 | Download | |
7 | gpt-4.5-preview-2025-02-27 | ✓ | 16.94% | $57.11 | 1 | Download | |
8 | gpt-4o-2024-05-13 | ✓ | 16.94% | $41.28 | 1 | Download | |
9 | gpt-4o-2024-05-13 | ✓ | 16.29% (-1.30/+1.30) | $25.62 (-0.10/+0.10) | 2 | Download | |
10 | gpt-4o-mini-2024-07-18 | ✓ | 15.42% (-0.43/+0.87) | $1.96 (-0.03/+0.02) | 3 | Download | |
11 | claude-3-5-sonnet-20241022 | ✓ | 15.31% | $2.75 | 1 | Download | |
12 | gpt-4o-2024-05-13 | ✓ | 15.31% | $21.01 | 1 | Download | |
13 | gpt-4o-mini-2024-07-18 | ✓ | 14.33% | $1.02 | 1 | Download | |
14 | gpt-4o-mini-2024-07-18 | ✓ | 13.36% | $1.33 | 1 | Download | |
15 | gpt-4o-2024-05-13 | ✓ | 13.36% | $13.38 | 1 | Download | |
16 | gpt-4o-mini-2024-07-18 | ✓ | 13.03% | $1.21 | 1 | Download | |
17 | gpt-4o-mini-2024-07-18 | ✓ | 12.05% | $0.76 | 1 | Download | |
18 | gpt-4o-mini-2024-07-18 | ✓ | 10.75% | $0.81 | 1 | Download | |
19 | gpt-4o-mini-2024-07-18 | ✓ | 9.88% (-0.76/+0.54) | $0.64 (-0.00/+0.00) | 3 | Download | |
20 | claude-3-5-haiku-20241022 | ✓ | 9.45% | $1.25 | 1 | Download | |
21 | gpt-4o-2024-05-13 | ✓ | 8.79% | $4.32 | 1 | Download | |
22 | gpt-4o-mini-2024-07-18 | ✓ | 8.47% | $1.64 | 1 | Download | |
23 | gpt-4o-mini-2024-07-18 | ✓ | 8.47% | $0.25 | 1 | Download | |
24 | gpt-4o-2024-05-13 | ✓ | 8.47% | $23.81 | 1 | Download | |
25 | Meta-Llama-3.3-70B-Instruct-Turbo | ✓ | 7.17% | $0.30 | 1 | Download | |
26 | gpt-4o-mini-2024-07-18 | ✓ | 6.51% | $1.07 | 1 | Download |
Accuracy vs. Cost Frontier for USACO
This plot shows the relationship between an agent's performance and its token cost. The Pareto frontier (dashed line) represents the current state-of-the-art trade-off. The error bars indicate min-max values across runs.
Heatmap for USACO
The heatmap visualizes success rates across tasks and agents. Colorscale shows the fraction of times a task was solved across reruns of the same agent. The "any agent" performance indicates the level of saturation of the benchmark and gives a sense of overall progress.
Failure Analysis (Experimental)
Select an agent to see a detailed breakdown of failure categories and their descriptions. This analysis helps understand common failure patterns and areas for improvement. Failure reports are usually available for the top 2 agents.
Failure Categories
Distribution of Failures
Additional Resources
Getting Started
Want to evaluate your agent on USACO? Follow our comprehensive guide to get started:
View DocumentationTask Details
Browse the complete list of USACO tasks, including difficulty levels and categories:
View Tasks