{"slug": "jetbrains-mellum2-1-from-2-to-47-on-swe-bench", "title": "JetBrains Mellum2.1: From 2% to 47% on SWE-Bench", "summary": "JetBrains released Mellum2.1, a 12B-parameter mixture-of-experts coding model that scores 47% on SWE-bench Verified, up from 2% for the prior version, according to byteiota. JetBrains trained Mellum2.1 with reinforcement learning as its main training loop, released it under the Apache 2.0 license, and the model runs on a 12 GB GPU.", "body_md": "JetBrains Mellum2.1 hits 47% on SWE-bench Verified using reinforcement learning as its main training loop. Apache 2.0, 12B MoE, runs on a 12 GB GPU. Here's how to use it.\n\nThe post \nJetBrains Mellum2.1: From 2% to 47% on SWE-Bench\n appeared first on \nbyteiota\n.", "url": "https://wpnews.pro/news/jetbrains-mellum2-1-from-2-to-47-on-swe-bench", "canonical_source": "https://byteiota.com/jetbrains-mellum2-1-from-2-to-47-on-swe-bench/", "published_at": "2026-10-31 01:10:39+00:00", "updated_at": "2026-10-31 01:19:31.210641+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research", "ai-tools", "machine-learning"], "entities": ["JetBrains", "Mellum2.1", "SWE-bench Verified", "byteiota"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/jetbrains-mellum2-1-from-2-to-47-on-swe-bench", "markdown": "https://wpnews.pro/news/jetbrains-mellum2-1-from-2-to-47-on-swe-bench.md", "text": "https://wpnews.pro/news/jetbrains-mellum2-1-from-2-to-47-on-swe-bench.txt", "jsonld": "https://wpnews.pro/news/jetbrains-mellum2-1-from-2-to-47-on-swe-bench.jsonld"}}