Technical benchmark suite
LiveBench
Release-based technical evaluation, with a newer long-horizon branch.
Key information
- Focus
- Technical question answering in LiveBench 1.0; long-horizon agent work in the site's newer 2.0 edition.
- Measure
- Release- and task-specific scores; editions must be distinguished.
- Version / scope
- LiveBench 2.0 site; LiveBench 1.0 retained separately
- Update pattern
- The original project describes recurring question releases. A site deployment or refreshed table is not proof of a new evaluation date.
- Publisher update
- Not consistently disclosed
- Editorial verification
- 15 October 2026
In plain English
What makes it different?
LiveBench was designed around refreshable tasks and checkable answers. Its original suite covers areas such as mathematics, reasoning, coding and instruction following. The current site presents LiveBench 2.0, while retaining the earlier leaderboard under a separate versioned route.
Sources and attribution are listed below.
What it measures
Technical question answering in LiveBench 1.0; long-horizon agent work in the site's newer 2.0 edition.
Strengths
- Dated question sets make the tested material identifiable.
- Task breakdowns expose differences hidden by a single overall score.
Limitations
- Fresh questions reduce some contamination risks; they cannot prove that all training overlap is absent.
- Version changes and different task coverage make cross-edition headline scores unsuitable for direct comparison.
How the evaluation works
In the original suite, answers are checked against task-specific ground truth, reducing reliance on a model judge. The release date selects the questions being tested.
The publisher's frontend repository distinguishes the 1.0 data flow from the 2.0 site. An older question set is not a measurement on the newer suite.
Configuration matters
Record task release, API model ID, reasoning effort and generation limits. Check whether a result is from the original question suite or an agentic edition before interpreting it.
Primary methods: Original evaluation repository and paper links ↗Publisher's version and frontend documentation ↗Original LiveBench paper ↗
Results at the publisher
Scores stay with the original source
We have not documented the permissions needed to reproduce this leaderboard here. You can still explore its methods and visit the publisher for results.
Versions & freshness
Verified 15 October 2026The live site's title and the publisher's frontend README identify the 2.0/1.0 split. A complete current 2.0 task release was not independently pinned in this editorial review.
Official site available as a JavaScript application; repository documentation accessible. The original project describes recurring question releases. A site deployment or refreshed table is not proof of a new evaluation date.
Our review date is not an evaluation date. Individual run dates are not consistently available; they remain unknown here. No automatic result collection is enabled.
Publication & reuse
The original evaluation code has an Apache 2.0 licence. A separate, sufficiently explicit permission for republishing the current leaderboard data and retaining its history has not been documented.
Apache 2.0 applies to the original evaluation code; it is not treated as a grant for the current leaderboard CSVs or the newer site.
View the scope of the reuse decision
- Result fields
- No result fields cleared
- Public result display
- not enabled
- Storage
- not enabled
- Historical copies
- not enabled
- Updates
- not enabled
- Commercial reuse
- not enabled
- Derived ratings
- No scoped grant documented; not enabled.
- Attribution
- Credit the named publisher and link the primary sources. Attribution alone does not grant reproduction rights.
This decision does not clear mixed-source scientific recommendations. Publisher content, software, datasets and model licences are separate. No partnership is implied.
Primary sources & attribution
This guide is original scor.es commentary about LiveBench research team's work. The links below support the identity, methods, versions and reuse review.
- Official website ↗Checked 15 October 2026
- Original evaluation repository and paper links ↗Checked 15 October 2026
- Publisher's version and frontend documentation ↗Checked 15 October 2026
- Original LiveBench paper ↗Checked 15 October 2026
Editorial edition 2026-10-15.1 · reuse review directory-rights-1. Re-check due 14 November 2026.