{
  "schema_version": "1.2.0",
  "rubric_version": "1.1",
  "license": "CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)",
  "record": {
    "id": "O-0212",
    "slug": "gensyn",
    "title": "Gensyn",
    "aliases": [],
    "status": "published",
    "last_reviewed": "2026-09-25",
    "review_interval_days": 90,
    "steward": null,
    "provenance": {
      "drafted_by": "ai",
      "reviewed_by": [
        "codex-review"
      ]
    },
    "risk_flags": [],
    "flags": [
      "provider-reported"
    ],
    "kind": "company",
    "homepage": "https://www.gensyn.ai/",
    "one_liner": "A company building AI systems that forecast events; developer of the Verde verification protocol and the REE runtime for reproducible model inference.",
    "sources": [
      {
        "source": "S-3560",
        "supports": "self-description; Delphi and REE as products",
        "locator": "homepage"
      },
      {
        "source": "S-1809",
        "supports": "Verde paper, most authors at Gensyn; dispute protocol and RepOps"
      },
      {
        "source": "S-1812",
        "supports": "REE: reproducible LLM inference with receipts; MIT-licensed SDK, proprietary binaries",
        "locator": "README"
      },
      {
        "source": "S-1810",
        "supports": "Verde and RepOps deployed in Judge (provider-reported)"
      },
      {
        "source": "S-1811",
        "supports": "Judge, an evaluation service built on Verde and REE (provider-reported)"
      },
      {
        "source": "S-3022",
        "supports": "Delphi live; markets settled inside REE produce re-runnable receipts (provider-reported)"
      }
    ],
    "type": "organization",
    "url": "https://trustbutveri.fyi/organizations/gensyn/",
    "source_file": "content/organizations/gensyn.md",
    "flags_all": [
      "provider-reported"
    ],
    "body_markdown": "Gensyn builds AI forecasting systems [[S-3560]]. Its verification work includes [[I-0015|Verde and RepOps]], which check machine-learning jobs delegated to compute providers [[S-1809]]:\n\n- Most of the authors of the paper that introduces Verde are at Gensyn [[S-1809]]. In Verde, several providers run the same machine-learning job, and a dispute between them is narrowed to a single operation that a referee re-runs [[S-1809]]. The paper's RepOps library fixes the order of floating-point operations so that results match across hardware [[S-1809]].\n- Gensyn publishes REE, its Reproducible Execution Environment, which it reports runs LLM inference reproducibly and writes receipts that let others re-run it [[S-1812]]. The SDK is MIT-licensed, and the compiler and operator binaries ship under a proprietary licence [[S-1812]]. See [[M-0002|deterministic and bit-exact inference]].\n- Gensyn reports that Verde and RepOps are deployed in Judge, its AI evaluation service [[S-1810]] [[S-1811]]. It reports that Delphi, its information-market app, is live, and that markets settled by open models inside REE produce receipts that anyone can re-run [[S-3022]].",
    "body_text": "Gensyn builds AI forecasting systems [S-3560]. Its verification work includes Verde and RepOps, which check machine-learning jobs delegated to compute providers [S-1809]: - Most of the authors of the paper that introduces Verde are at Gensyn [S-1809]. In Verde, several providers run the same machine-learning job, and a dispute between them is narrowed to a single operation that a referee re-runs [S-1809]. The paper's RepOps library fixes the order of floating-point operations so that results match across hardware [S-1809]. - Gensyn publishes REE, its Reproducible Execution Environment, which it reports runs LLM inference reproducibly and writes receipts that let others re-run it [S-1812]. The SDK is MIT-licensed, and the compiler and operator binaries ship under a proprietary licence [S-1812]. See deterministic and bit-exact inference. - Gensyn reports that Verde and RepOps are deployed in Judge, its AI evaluation service [S-1810] [S-1811]. It reports that Delphi, its information-market app, is live, and that markets settled by open models inside REE produce receipts that anyone can re-run [S-3022].",
    "referenced_by": [
      {
        "id": "I-0015",
        "title": "Verde and RepOps (Gensyn)",
        "url": "https://trustbutveri.fyi/implementations/gensyn-verde-repops/"
      },
      {
        "id": "S-1809",
        "title": "Verde: Verification via Refereed Delegation for Machine Learning Programs",
        "url": "https://trustbutveri.fyi/sources/arun-verde-refereed-delegation/"
      },
      {
        "id": "S-3022",
        "title": "Building Delphi: Pricing, Settlement, and Agentic Trading",
        "url": "https://trustbutveri.fyi/sources/gensyn-building-delphi/"
      },
      {
        "id": "S-3560",
        "title": "Gensyn: machines that predict the future",
        "url": "https://trustbutveri.fyi/sources/gensyn-homepage/"
      },
      {
        "id": "S-1811",
        "title": "Introducing Judge",
        "url": "https://trustbutveri.fyi/sources/gensyn-introducing-judge/"
      },
      {
        "id": "S-1812",
        "title": "gensyn-ai/ree: Gensyn Reproducible Execution Environment (GitHub repository)",
        "url": "https://trustbutveri.fyi/sources/gensyn-ree-code/"
      },
      {
        "id": "S-3023",
        "title": "Reproducible Execution Environment (REE) (Gensyn documentation)",
        "url": "https://trustbutveri.fyi/sources/gensyn-ree-docs/"
      },
      {
        "id": "S-1810",
        "title": "Verde Verification System In Production",
        "url": "https://trustbutveri.fyi/sources/gensyn-verde-in-production/"
      }
    ]
  }
}