Eragon(@EragonAI):Today we are publishing our first collaborative research with @Eon_io_ We gave an Eragon agent a readable guide to a company's systems, then let GEPA rewrite that guide from the agent’s own failed traces. GPT-5.6 Luna is the cheapest model we tested, and cold it was the weakest of the GPT and Claude models: 7/35. Post GEPA: 29/35.

2026.10.08 17:31

Today we are publishing our first collaborative research with @Eon_io_ We gave an Eragon agent a readable guide to a company's systems, then let GEPA rewrite that guide from the agent’s own failed traces. GPT-5.6 Luna is the cheapest model we tested, and cold it was the weakest of the GPT and Claude models: 7/35. Post GEPA: 29/35.

显示更多