Benchmark Almanac

Beijing University of Posts

We track 8 benchmarks that Beijing University of Posts made or helped make.

0 in use 7 with public code 4 with a leaderboard here

Benchmarks

Name Kind Released Best score Status
OJBench Models write C++ or Python solutions to 232 competition problems from NOI and ICPC. Judge test cases score each solution by pass@1. Competitive programming Jun 2025 70% Fading
We-Math Models solve visual math problems decomposed into sub-problems by knowledge concept. Scoring compares sub-problem and full-problem results to diagnose knowledge and generalization gaps. Visual math and diagrams Jul 2024 89% Fading
FinanceReasoning Models solve financial numerical reasoning problems that need one or more finance formulas. Answers are numbers checked against reference Python solutions. Finance Jun 2025 86.9% System cards only
SWE-Game Agents build, complete, repair, or port games across 247 tasks derived from 41 reference games. Runtime probes and input replays check gameplay mechanics and preserved behavior. Vision-language rubrics separately grade presentation in construction tasks. Building whole projects Sep 2026 64.0 Not in cards
CoralVQA Models answer questions about coral identity, growth, and health in reef photographs. GPT-4o checks answers against references across 16 question types. General visual questions May 2025 41.8% Not in cards
Animal-Bench Models answer 41,839 questions about animal videos. The overall score averages accuracy across 13 tasks, including action prediction and recognizing predator and prey behavior. Video understanding Not in cards
LlamaTouch Agents carry out requests in Android apps. The evaluator checks whether each run visited the required app or system states. Phones Apr 2024 Not in cards
SpatialGenEval Text-to-image models generate scenes from 1,230 prompts, and vision-language judges answer 12,300 spatial questions about the resulting images. Image generation and editing Jan 2026 Not in cards