Prewire · case file · public copy of the record
FILE P367 PREWIRE receipts archive the live wire
CATCH #2167 DETECTED OCT 19, 2026 22:19 UTC
THE CATCHscore 7/10
EXHIBIT 2167 · URGENT
Oct 19, 2026 22:19 UTC github_org github.com/openai
OpenAI made a new public repo called MultiLangBench, an evaluation benchmark for language models across multiple languages using SQLite.

READY TO POST:

openai just open-sourced multilanguagebench! looks like they're releasing their eval for benchmarking llms across different languages.

interesting to see openai drop multilanguagebench on github. definitely hints at their focus on multilingual llm performance with a sqlite-based eval.

EVIDENCE — the receipts
openai: 1 new public repo

  + openai/MultiLangBench — An eval based on SQLite that benchmarks language models in multiple languages.

caught by the 20-minute sweep · the live wire · all receipts

THE WIRE RECEIPTS ACCESS @VEDOLOS TERMS PRIVACY break the story before it breaks