Hex Platform Releases Native LLM Evals with Model Sweeps and Git Tracking
Madisonkanna · x · 2026-08-06
Data platform Hex has officially launched Evals, a native evaluation feature built on the team's two years of experience developing internal testing tools. It aims to help developers better test knowledge base agents.
Key features include:
- LLM-as-a-judge rubrics: Allows for nuanced grading based on specific criteria.
- Model sweeps: Enables developers to easily test and compare new models as they are released.
- Code-defined workflows: Everything is defined in code, making agent-based setup seamless and allowing all changes to be tracked via Git version control.
More from coding & agent
- Using AI to Discover Business Rules in Legacy Systems: Experiences & Challenges — CF7_Gaming · 2026-08-06
- Frontier Models Tested on Complex Agents: DeepSeek Wins on Cost Despite Inefficiency — rohanpaul_ai · 2026-08-06
- Context Engineering: Managing Agent Sources with a Dynamic Trust Curve — anselm · 2026-08-06
- Multi-Model Orchestration Workflow Beats ElevenLabs in Enterprise AI Dubbing — markjeffrey · 2026-08-06
- Keyless API Calls Bypass Zero Data Retention, Exposing Privacy Flaws — kleffew94 · 2026-08-06
- Developer Showcases Agentic IDE That Builds Itself — jasonkneen · 2026-08-06