LongTail-SWE: 202 real tasks across 13 underrepresented languages, Qwen3.8 solves 43% vs Qwen3.6's 17%
Fabulous_Pollution10 · reddit · 2026-11-10
A team released LongTail-SWE (longtail-swe.com), a benchmark of 202 real GitHub tasks in 13 languages rarely covered by coding benchmarks: Elixir, OCaml, Haskell, Clojure, Kotlin, Swift, Julia, R and more. On a limited compute budget they compared 27B open models head-to-head: Qwen3.8 solves 43% of tasks versus 17% for Qwen3.6. The leaderboard also includes Kimi K3, DeepSeek, GLM and closed models with per-language breakdowns. Caveat: Qwen3.8 was used to select hard tasks, so its score is biased slightly down.
Related event: Nebius Releases LongTail-SWE Benchmark for 13 Underrepresented Languages(3 posts)→
More from coding & agent
- Bolt: Gemma 4 runs 100% on-device on iPhone to drive a robot dog health companion — DynamicWebPaige · 2026-11-11
- Dev shows off live voice-call AI agent 'Elise' that answers family calls and triages urgency — bigaiguy · 2026-11-11
- agents.md isn't just for AI agents — humans can grok codebases fast with it too — k7agar · 2026-11-11
- Animate real places in 3D with Claude and free map data from AWS and Overture — drbronk · 2026-11-11
- Microsoft Foundry pattern: version and deploy Toolboxes separately from agents behind stable endpoints — WirelessLife · 2026-11-11
- Ethan Mollick lets a swarm of agents argue their way into building a fantasy transport sim — emollick · 2026-11-11