The Economics of Context: Cost-Optimal Prompt Management for Long-Running LLM Agents
synthfoundry.substack.com/p/…
1
177
In my case, per task Terra (6.1) costs more. Tokens were attributed to each mission from the brief, and turned into weekly-quota points with weights fitted over 45 hours of quota readings (R² 0.82): 3.4 points per million output tokens, 0.17 per million new input tokens, 0.015 per million cached input tokens. The fit, made mostly on Sol hours, predicted Terra’s two Terra-only hours within half a point, so Terra is not priced higher per token.
🚨 GPT 6.1 Terra Is The Cheapest Frontier Model Per Task
- cheaper than other frontier models
- comparable to Argon and Sonnet in total score
- very good at browser use
A really good fit for personal agents and as a chatbot
1
135















