Which AI Model Should You Use for Which Task?
Summary withheld: an Opus review found that the machine-written summary did not match the post. Read the original post.
Read it hereOpen in Claudehttps-freddysblog-com-2026-08-23-which-model-should-you-use-for-which-task--faa80c64bd.md
post Freddys blog
"A good harness manufactures verifiability the raw task didn't have - and every bit of verifiability you add lets you reach for a weaker model."
Quotes
- “A good harness manufactures verifiability the raw task didn’t have - and every bit of verifiability you add lets you reach for a weaker model.” (Shows that the scaffolding around a model is as important as the model itself in determining which tier you need)
- “Staff each task with the cheapest model that clears that bar, and save your architects for the handful where judgment really is the product.” (Encapsulates the economic principle: use strong judgment where it adds value, automate the rest)
Context
- Features: model selection framework, task verifiability assessment, harness and scaffolding, skills as runbooks, automatic routing, open-weights models, self-hosted models, model comparison
Source: Freddys blog, community blog. Summary, key points and quotes are derived (CONTENT-NOTICE.md); read the original for the full text.