Harvey · 2× · last seen Aug 19, 2026
45
Momentum
Harvey Tenet is Harvey's first in-house, proprietary AI model, post-trained end-to-end for legal reasoning. Previously, Harvey relied solely on third-party general-purpose foundation models (including OpenAI, Anthropic, Google); Tenet was trained using Harvey's own version of the open-weight Kimi K3 model on mock disputes and case files. According to Harvey, Tenet achieves frontier-level performance on prominent legal benchmarks on par with the strongest general models, but at open-source cost, making continuous agent deployment across every matter practical. The model is part of the broader "Harvey II" release and is intended to lay the groundwork for firm-specific specialized models in the future.
Momentum trend
21.05.19.08.
Features
| Compliance/Certification | SOC 2 Type II, ISO 27001, ISO 27701, ISO 42001, CCPA, GDPR |
| Deployment Model | Cloud-based platform with enterprise security controls (SAML SSO, audit logs, IP allow-listing); model built on open-weight technology (Kimi K3) |
| Use Case Scope | Legal reasoning for law firms and in-house legal teams: contract analysis, due diligence, compliance, litigation, M&A reviews |
| Integrations | iManage, NetDocuments, SharePoint, Google Drive, Aderant, Ironclad, APIs, 500+ regional knowledge sources |
| Platform | Integrated into the Harvey platform (Agents, Assistant, Vault, Knowledge, etc.) |
| Release Date | Announced as part of "Harvey II" (blog post ~August 19, 2026) |