Harvey and Applied Compute train legal-specific model with SOTA accuracy at fraction of cost
rhythmrg · x · 2026-08-15
Applied Compute partnered with legal AI company Harvey to train an open-source model for Review Table, one of Harvey's highest-volume products. The model achieves state-of-the-art accuracy at a fraction of the cost and latency of frontier alternatives. Training for production required building a representative eval and understanding real lawyer workflows.
Related event: Harvey Post-trains GLM-5.2 to Cut Legal AI Costs by 50%(4 posts)→
More from Companies & People
- OpenAI annualized revenue run rate doubles to over $40B — Polymarket · 2026-08-15
- Rumor: Anthropic uses internal model better than Mythos 5 — kimmonismus · 2026-08-15
- Lesson from frontier labs: Ship blog posts or ship work — dejavucoder · 2026-08-15
- YC-backed Stratum builds AI agents to automate government workflows — ycombinator · 2026-08-15
- Grok Bot's Clean Slate Could Be Its Advantage: A Path to the Consumer ChatGPT Moment for Personal Agents — manosaie · 2026-08-15
- Analysis: Cybersecurity budget becomes the gatekeeper for enterprise AI — BenBajarin · 2026-08-15