MSbench
Thesis catalogue (v3)coming soon Automated catalogue Skills Sources Known issues Downloads & APIcoming soon About
DMSL @ UCY
Preview · the database itself is not public yet · data shown as built, issues documented on Known issues

Many LLMs,
scored in one place.

Benchmark results for language models, collected for LLM-MS: a decentralized edge system that finds the right models across a peer-to-peer network. Its Index ranks models using these scores. Every score is traced back to the public leaderboard it came from.

Where the data comes from Automated catalogue
15,208models
51leaderboards
591,502scores
1,261tests

Two catalogues

Thesis catalogue (v3)

Thesis versionFrozen 23 Oct 2026

Built once from 51 leaderboards. Raw scores, comparable only within one test.

15,208 models · 591,502 scores · 51 sources

Browse the dataComing soon

Automated catalogue

Automated versionRelease 2026-11-08

Re-pulled from live leaderboards, with the review's fixes enforced as build checks.

5,443 models · 74,644 scores · 13 sources

Browse the dataComing soon

Where it stands

  • Where the data comes from: public leaderboards, each with its own licence. See Sources.
  • Rankings are provisional: a rank needs only 2 tests from 2 leaderboards, some sources are stale, and the review queue is undecided. See Skills and Known issues.
  • Preview: the database is not published until the work is, and each licence is checked. See About.
DMSL @ UCY
MSbench, the LLM-MS capability catalogue · Data Management Systems Laboratory (DMSL), University of Cyprus · Scores belong to their leaderboards, see Sources
Powered by Datasette 0.65.5