BytePort · public benchmark node

A source-linked view of public model benchmarks.

23 objective tasks across 7 categories. Choose a frozen release, compare category scores, and inspect every published subtask.

Release2026-06-25
2024-06-24 · v12026-06-25 · latest

Showing the latest indexed release.

01

Leaderboard

Overall and category averages are sortable. Select one category to reveal its subtasks; open any model row for the complete score profile.

Models—published rows
Tasks—objective checks
Categories—equal-weight groups
Top score—current leader
Snapshot—frozen source
Loading release…

Scores are calculated from the selected LiveBench public table. BP Eval is an independent derivative index and is not affiliated with LiveBench or model providers.

02

Insights

Capability and efficiency views for the selected release.

Quality vs. cost

Published cost per successful task, plotted against overall score.

Category profile

The three leading snapshots compared across category averages.

03

About this index

Frozen releases

Dates remain attached to every table so later updates do not silently rewrite an earlier comparison.

Visible aggregation

Category values are task means; overall is the arithmetic mean of the displayed categories.

Independent presentation

Public results are reorganized here for inspection. Evaluation ownership remains with the primary source.