Mozilla draft report puts the open-model lag at about four months

single source· 1 articles · confidence: low · first seen 2026-09-15 12:00 UTC

What this means for you

Nothing to act on yet: no published report, no benchmark list and no evaluation date, so the four-month and five-times figures cannot be checked. If you are weighing open weights against a paid API, wait for the method before you move.

Mozilla has a report in preparation comparing cheap open-weight models with the paid frontier, and Ars Technica has previewed it ahead of publication. On Ars's account, the paid models hold roughly a four-month lead in capability while costing about five times as much, with open Chinese models named as the closest challengers. No model names, benchmark list or evaluation date appear in the material available, and the report is not yet out, so none of the figures can be checked against a published method.

Key facts

  • ·Mozilla has a report in preparation on how far cheap open-weight models have closed the gap with paid frontier models, previewed by Ars Technica before publication. source
  • ·Ars Technica's preview puts the paid frontier models' capability lead at about four months. source
  • ·The same preview puts the price of that lead at roughly five times the cost. source
  • ·Open Chinese models are named as the closest challengers to the paid frontier. source
  • ·No model names, benchmark list or evaluation date are given in the material available. source

What the sources say

  • Ars Technica AI — Previews an unpublished Mozilla report comparing cheap open-weight systems against paid frontier models.

Sources

The original reporting. Follow these — they did the work.

← the wire