OpenCode Models Comparison

This article provides a detailed comparison of AI models for cost-efficiency, specifically within the OpenCode platform. It introduces a practical filter-then-rank decision procedure: set minimum capability thresholds, filter out under-qualified models, rank by effective cost ascending, and pick the cheapest. Effective cost is calculated using a blended per-million-token rate accounting for cache hits and input/output token ratios. The article includes a table with sliders for Intelligence, Coding, and Agentic scores, and discusses capability tiers for various tasks like simple classification, complex codegen, and autonomous agents. Observations on specific models like Claude Sonnet 5 highlight cost inefficiencies. This is a technical guide for developers choosing AI models.