Longtail-SWE: benchmark for reasoning in Underrepresented Languages

First reported Nov 11, 2026, 10:59 UTC

Source reports

Longtail-SWE: benchmark for reasoning in Underrepresented Languages by Nebius team main insight: “agent’s apparent coding failure can actually be a spec or test issue” also, ai agents are surprisingly good at Erlang and Haskell, even 🇨🇳 ones https://arxiv.org/pdf/2611.10304 — Закрепи.sender, Nov 11, 2026, 10:59 UTC
Longtail-SWE: benchmark for reasoning in Underrepresented Languages by Nebius team main insight: “agent’s apparent coding failure can actually be a spec or test issue” also, ai agents are surprisingly good at Erlang and Haskell, even 🇨🇳 ones https://arxiv.org/pdf/2611.10304 — Rickler Feed, Nov 11, 2026, 10:59 UTC

Excerpts are quoted from the linked sources, which own their text.