4DCodeBench · Evaluation viewer
Visual Evaluation
Reconstruction quality, run status, and API cost, side by side on one timeline.
0
scenarios
Observation & reconstruction
Case overview
Iterative agents
Agentic Harness
0 traces
Single API call
Zero-shot
0 traces
Iterative refinement
Multiround
0 runs
Each card previews the final round and reports cumulative metrics; expand it to compare the reference and all rounds.
4DCodeBench · Model gallery
Sample Gallery
Browse runs across datasets, then inspect any sample side by side.
0
runs
Cross-dataset preview