{"slug": "adversarial-speculative-reasoning-improves-long-horizon-agents", "title": "Adversarial Speculative Reasoning Improves Long Horizon Agents", "summary": "A paper submitted to arXiv on 3 November 2026 reports that training a draft model to be rejected at critical decision points in speculative decoding raises end-to-end success rates on complex agentic coding benchmarks without changing the target model or its output distribution. The authors found that numerical noise grows within verification windows and that draft acceptance results can shift the output distribution of frontier open-weight models after 30 steps, moving long-horizon reasoning accuracy in random directions. Their adversarial speculative reasoning approach trains only 0.06% to 0.28% of hosted parameters, recasting speculative decoding from a pure accelerator into an active allocator for context precision.", "body_md": "# Computer Science > Artificial Intelligence\n\n  [Submitted on 3 Nov 2026]\n\n# Title:Adversarial Speculative Reasoning Improves Long Horizon Agents\n\n[View PDF](http://arxiv.org/pdf/2611.05345v1)\n\n[HTML (experimental)](https://arxiv.org/html/2611.05345v1)\n\nAbstract:Speculative decoding has been treated purely as a systems optimization that speeds up large language model decoding without changing the model behavior. Under this view, the growing effort to train better draft models has been focused solely on acceptance rate. We argue that this view does not hold true for long-horizon agent reasoning. Measuring frontier open-weight models on complex agentic tasks with different speculative decoding approaches, we found that numerical noise grows within verification windows, and the draft acceptance results of the thinking process can change the output distribution of the model after 30 steps, thus noticeably influencing the reasoning accuracy of long-horizon agents in random directions. In this work, we improve the reasoning capability of target models by controlling such noise. While we cannot change the overall token distribution verified by the target model, we can encode critical reasoning with highest precision by training the draft model to be rejected. We identify these points in agent traces and train the draft model to favor adversarial preference. At inference the trained draft and the target part ways precisely at these decision points, verification restarts, so that critical thinking tokens are sampled from the target model and encoded under its least noisy conditions. This approach makes critical reasoning better indexed and more durable over a long horizon. On complex agentic coding benchmarks, our approach raises end-to-end success rate with no change to the target model or its output distribution. Our results recast the speculative decoding mechanism from a pure accelerator into an active allocator for context precision that improves the accuracy of long-horizon agents by training 0.06% to 0.28% of hosted parameters.\n    \n\n### Additional Features\n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer \n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers \n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps \n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations \n\n*(*[What are Smart Citations?](https://www.scite.ai/))\n# Code, Data and Media Associated with this Article\n\nalphaXiv \n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers \n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub \n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub \n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face \n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast \n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))\n# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower \n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender \n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))\n# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.\n\nHave an idea for a project that will add value for arXiv's community? [**Learn more about arXivLabs**](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/adversarial-speculative-reasoning-improves-long-horizon-agents", "canonical_source": "http://arxiv.org/abs/2611.05345v1", "published_at": "2026-11-04 19:26:25+00:00", "updated_at": "2026-11-04 19:53:05.445428+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-research", "machine-learning"], "entities": ["arXiv", "Adversarial Speculative Reasoning"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/adversarial-speculative-reasoning-improves-long-horizon-agents", "markdown": "https://wpnews.pro/news/adversarial-speculative-reasoning-improves-long-horizon-agents.md", "text": "https://wpnews.pro/news/adversarial-speculative-reasoning-improves-long-horizon-agents.txt", "jsonld": "https://wpnews.pro/news/adversarial-speculative-reasoning-improves-long-horizon-agents.jsonld"}}