LiveCodeBench
Contamination-free live benchmark for code generation
💸 No earnings reported yet
What it is
LiveCodeBench continuously collects new competitive-programming problems as they are published, so models are scored on problems released after their training cutoff. Beyond code generation it evaluates self-repair, execution prediction, and test-output prediction, and publishes a rolling leaderboard used to compare coding models without the benchmark contamination that plagues static suites.
How AI plugs in
Alternatives & related tools
Artificial Analysis
Independent AI model benchmarking
Epoch AI
Research and benchmarks tracking AI progress
Apollo Research
AI deception and scheming evaluation lab
LMArena
Crowdsourced human-preference model leaderboard
METR
Independent frontier-model dangerous-capability evaluator
MLCommons
Open engineering consortium behind MLPerf and AI safety benchmarks
★ Reviews
No reviews yet — be the first.Your rating