๐Ÿ“„Stalecollected in 17h

Independent GPT-OSS-20B Benchmark Reproduction

Independent GPT-OSS-20B Benchmark Reproduction
PostLinkedIn
๐Ÿ“„Read original on ArXiv AI

๐Ÿ’กFirst independent repro of OpenAI gpt-oss-20b agent benchmarks + open harness

โšก 30-Second TL;DR

What Changed

Reverse-engineered tools from gpt-oss-20b training distribution via prompts

Why It Matters

Validates OpenAI's gpt-oss-20b agent claims independently, boosting trust in published benchmarks. Provides open tools for community agent development and evaluation.

What To Do Next

Clone https://github.com/borislavmavrin/harmonyagent.git and test gpt-oss-20b on SWE-bench.

Who should care:Researchers & Academics

Key Points

  • โ€ขReverse-engineered tools from gpt-oss-20b training distribution via prompts
  • โ€ขBuilt harmonyagent GitHub repo for native message encoding
  • โ€ขReproduced OpenAI scores: SWE Verified HIGH 60.4% (pub. 60.7%), MEDIUM 53.3% (53.2%)
  • โ€ขAchieved 91.7% on AIME25 with tools (pub. 90.4%)
๐Ÿ“ฐ

Weekly AI Recap

Read this week's curated digest of top AI events โ†’

๐Ÿ‘‰Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: ArXiv AI โ†—