{
  "id": 80926,
  "title": "China’s DeepSeek beefs up agentic AI with ‘harness’ tests as V4 model jolts Silicon Valley",
  "url": "https://urgent.news/2026/08/03/chinas-deepseek-beefs-up-agentic-ai-with-harness-tests-as-v4-model",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-03T11:00:07.000Z",
  "source": {
    "name": "South China Morning Post",
    "slug": "south-china-morning-post",
    "url": "https://www.scmp.com/tech/tech-trends/article/3362792/chinas-deepseek-beefs-agentic-ai-harness-tests-v4-model-jolts-silicon-valley"
  },
  "original_language": "en",
  "account": "Chinese AI firm DeepSeek is intensifying its foray into agentic AI technology by inviting open-source developers to beta test its forthcoming \"harness\" software, according to a social media update by Cui Tianyi, who leads the harness team. The software aims to transform large language models, such as DeepSeek's latest V4 Flash model, into autonomous agents capable of executing multi-step tasks and complex workflows. Agentic frameworks have become a focal point in the AI industry following Anthropic's successful release of Claude Code.\n\nDeepSeek's Cui Tianyi, who joined the company in March to spearhead the harness project, stated that while the team harbored ambitious goals, it was currently understaffed and actively recruiting across multiple platforms. The agentic AI push coincides with DeepSeek's CEO and co-founder, Liang Wenfeng, focusing on cost-effective, high-capacity models as a strategy towards achieving artificial general intelligence (AGI). This approach has sparked concerns in Washington and Silicon Valley about the growing influence of Chinese open-weight models.\n\nRecently, DeepSeek unveiled \"V4 Flash 0731\", a 284-billion-parameter lightweight iteration of its V4 series, which garnered praise for its performance comparable to OpenAI's lightweight GPT-5.6 Luna at a significantly lower cost. The new model is priced at US$0.03 per task, making it approximately 105 times cheaper than Anthropic's Claude Fable 5, which costs US$3.15 per task. This pricing advantage has prompted remarks such as \"Why pay $1 when I can pay $105\" from venture capitalist Chamath Palihapitiya.",
  "summary": "Chinese AI company DeepSeek is inviting open-source developers to test its \"harness\" software, designed to turn large language models into AI agents. This move is part of DeepSeek's push into agentic tech, accelerated by the release of its latest V4 Flash model. According to the South China Morning Post, DeepSeek is looking for open-source project developers to join beta testing.\n\nThe V4 Flash model has sent a cost-efficiency shock wave through Silicon Valley, with its pricing significantly lower than competitors. According to Artificial Analysis, a research firm, DeepSeek's V4-Flash charges $0.14 per million input tokens and $0.28 per million output tokens. This translates to an average cost of 3 cents per test, compared to 86 cents for Moonshot AI's Kimi K3, $1.86 for OpenAI's GPT-5.6 Sol, and $3.15 for Anthropic's Claude Fable 5, as reported by Free Malaysia Today and Economic Times Tech.\n\nThe developments highlight the rapid advancement of artificial intelligence by Chinese tech firms, with companies like Alibaba and DeepSeek aiming to gain traction among global developers. Chinese open-weight models, like those from DeepSeek and Moonshot AI, are becoming increasingly powerful and cost-effective, sparking a debate in the US tech industry about their adoption and potential security risks, as noted by Rest of World Tech and Channel News Asia.",
  "key_points": [
    "DeepSeek developing agentic AI with \"harness\" software for open-source beta testing.",
    "V4 Flash 0731 model priced at $0.03 per task, 105 times cheaper than Claude Fable 5.",
    "Agentic AI push aligns with DeepSeek's strategy for cost-effective, high-capacity models."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 4,
    "also_reported_by": [
      {
        "outlet": "Channel News Asia",
        "title": "DeepSeek's new AI model is by far the cheapest of well-known models to run, research firm says",
        "url": "https://urgent.news/2026/08/03/deepseeks-new-ai-model-is-by-far-the-cheapest-of-well-known-models-to",
        "published": "2026-08-03T05:41:43.000Z"
      },
      {
        "outlet": "Economic Times Tech",
        "title": "DeepSeek's new AI model is by far the cheapest of well-known models to run, research firm says",
        "url": "https://urgent.news/2026/08/03/deepseeks-new-ai-model-is-by-far-the-cheapest-of-well-known-models-to-72063",
        "published": "2026-08-03T05:43:40.000Z"
      },
      {
        "outlet": "SCMP Tech",
        "title": "China’s DeepSeek beefs up agentic AI with ‘harness’ tests as V4 model jolts Silicon Valley",
        "url": "https://urgent.news/2026/08/03/chinas-deepseek-beefs-up-agentic-ai-with-harness-tests-as-v4-model-84641",
        "published": "2026-08-03T11:00:07.000Z"
      }
    ]
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}