{"slug": "cohere-s-30b-coding-agent-beats-models-4x-its-size-on-one-h100-and-it-shouldn-t", "title": "Cohere's 30B Coding Agent Beats Models 4x Its Size on One H100 — and It Shouldn't", "summary": "Cohere's 30-billion-parameter coding agent, with only 3 billion active parameters, achieved a score of 0.6 on a benchmark, outperforming models four times its size while running on a single H100 GPU.", "body_md": "A 30-billion-parameter model with only 3 billion active parameters has no business landing 0.6\nContinue reading on Towards AI »", "url": "https://wpnews.pro/news/cohere-s-30b-coding-agent-beats-models-4x-its-size-on-one-h100-and-it-shouldn-t", "canonical_source": "https://pub.towardsai.net/coheres-30b-coding-agent-beats-models-4x-its-size-on-one-h100-and-it-shouldn-t-a61b62b4de72?source=rss----98111c9905da---4", "published_at": "2026-06-21 05:52:05+00:00", "updated_at": "2026-06-21 06:11:49.469362+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "ai-products"], "entities": ["Cohere", "H100"], "alternates": {"html": "https://wpnews.pro/news/cohere-s-30b-coding-agent-beats-models-4x-its-size-on-one-h100-and-it-shouldn-t", "markdown": "https://wpnews.pro/news/cohere-s-30b-coding-agent-beats-models-4x-its-size-on-one-h100-and-it-shouldn-t.md", "text": "https://wpnews.pro/news/cohere-s-30b-coding-agent-beats-models-4x-its-size-on-one-h100-and-it-shouldn-t.txt", "jsonld": "https://wpnews.pro/news/cohere-s-30b-coding-agent-beats-models-4x-its-size-on-one-h100-and-it-shouldn-t.jsonld"}}