{"slug": "deepseek-v4-flash-vs-glm-5-2-the-1-7-point-win-collapses-when-you-swap-the", "title": "DeepSeek V4-Flash vs GLM-5.2: The 1.7-Point Win Collapses When You Swap the Harness", "summary": "DeepSeek's July 31, 2026 release of DeepSeek-V4-Flash-0731, a 284-billion-parameter model with 13 billion active parameters, claims a 1.7-point lead over Z.ai's GLM-5.2 on Terminal-Bench 2.1 (82.7 vs 81.0), but the score was produced using an 'upcoming' DeepSeek Harness (minimal mode) framework that had not shipped, and documented harness-only effects as small as 2.1 points can erase that margin.", "body_md": "Member-only story\n\n# DeepSeek V4-Flash vs GLM-5.2: The 1.7-Point Win Collapses When You Swap the Harness\n\n**DeepSeek’s own chart says 82.7 on Terminal-Bench 2.1. Artificial Analysis measured 79. That 3.7-point gap is 2.2x the margin the 82.7 was defending.**\n\nOn July 31, 2026, DeepSeek shipped `DeepSeek-V4-Flash-0731`\n\nand published a nine-row agent benchmark chart. The chart is genuinely impressive: a 284-billion-parameter model with **13 billion active** at inference beats Z.ai's GLM-5.2 on every single row where both were measured, and comes within a few points of Claude Opus 4.8 on several.\n\nThen I read the footnote under the table.\n\nFor public Code Agent tasks, V4-Flash was tested using our upcoming DeepSeek Harness (minimal mode) framework. Settings: max tier, topp=0.95, temperature=1.0.\n\n**Upcoming.** The framework that produced every number on that chart had not shipped on the day the chart shipped. No one outside DeepSeek could run it.\n\nThat would be a footnote-sized problem if agent benchmark scores were harness-independent. They are not. The margin DeepSeek is claiming on Terminal-Bench 2.1 is **1.7 points** (82.7 vs GLM-5.2’s 81.0). The smallest harness-only effect I could find documented anywhere in the 2026 literature — adding one search subagent to an otherwise identical scaffold — is **2.1 points**. The floor of the…", "url": "https://wpnews.pro/news/deepseek-v4-flash-vs-glm-5-2-the-1-7-point-win-collapses-when-you-swap-the", "canonical_source": "https://pub.towardsai.net/deepseek-v4-flash-vs-glm-5-2-the-1-7-point-win-collapses-when-you-swap-the-harness-aa3327c87e26?source=rss----98111c9905da---4", "published_at": "2026-08-02 16:07:28+00:00", "updated_at": "2026-08-02 16:52:20.604919+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "ai-products"], "entities": ["DeepSeek", "DeepSeek-V4-Flash-0731", "Z.ai", "GLM-5.2", "Claude Opus 4.8", "Terminal-Bench 2.1", "DeepSeek Harness"], "alternates": {"html": "https://wpnews.pro/news/deepseek-v4-flash-vs-glm-5-2-the-1-7-point-win-collapses-when-you-swap-the", "markdown": "https://wpnews.pro/news/deepseek-v4-flash-vs-glm-5-2-the-1-7-point-win-collapses-when-you-swap-the.md", "text": "https://wpnews.pro/news/deepseek-v4-flash-vs-glm-5-2-the-1-7-point-win-collapses-when-you-swap-the.txt", "jsonld": "https://wpnews.pro/news/deepseek-v4-flash-vs-glm-5-2-the-1-7-point-win-collapses-when-you-swap-the.jsonld"}}