Tag archive
Claude API prices per million tokens, October 2026: Haiku 5.5 from $0.10 input, Sonnet 5.5 $2, Opus 5.5 $4, Fable 5.1 $10. Worked examples, caching and batch.
Claude Max API credits: $100 on Max 5x, $200 on Max 20x, up to $500 on Team. What $200 buys per model, what draws on it, and what does not.
A developer's look at Claude Opus 5: the effort parameter, thinking-on-by-default, the 1M-token context window, cache-safe tool changes, and how fallback routing handles safety-classifier refusals in production.
Claude's cache needs a 512–4,096 token prefix depending on model — miss it and nothing errors, nothing caches. Here's how to verify and fix
A wire-format comparison of Claude, OpenAI, and Gemini SSE streaming: event shapes, tool-call reassembly from partial JSON, and what to do when the stream dies mid-response.
Build an AI agent in Go from scratch: a tool-calling loop against the Anthropic API in ~120 lines, no frameworks, with real code and the gotchas I hit.
Claude doesn't actually use traditional code hooks — discover the 8-step pipeline that powers every tool call, from edge routing to result injection.