Tag archive
When your LLM bill moves, three separate things could have caused it. Your usage changed, your pricing changed, or the way your app talks to the model changed. Mixing them up is how teams end up thanking the wrong lever. A simple framework for pulling them apart.
Token savings are easy to market and hard to bank. Here’s the benchmark framework I trust for RTK-style compression: cost-per-success, tool-call failures, and latency.
1.25x cache writes start billing on 21 August 2026: OpenAI's new dashboard, Azure's missing...
Claude mid-conversation tool changes end a 12.5x cache-rewrite penalty in 2026 Summary. On...
Cut LLM spend 40-70% in 2026: an AI gateway and model-routing service Summary. The fastest...
Agentic RAG costs 2.5x classic RAG: the 2026 routing decision Summary. Agentic RAG is not...
Claude Opus 5 for coding agents: index 61 at half Fable 5's cost (2026) Summary. Claude...
Measured across GPT-5.6, Claude Fable 5, Qwen3.7-max and five more families: reasoning dominates the output bill. How to read every usage field, and cap them.
Vendors publish a prompt-cache token minimum. Measured across LLM families, auto-cache needs 1.4–2.4x more than the docs say; Claude's explicit cache is exact.
LLM output prices fell about 94% since 2023: how to cut your AI bill without losing quality...
Kimi K3 for coding teams in 2026: benchmarks, real cost and adopt-vs-wait Summary. On July...
A spreadsheet-ready expected-cost model for agent workflows that includes retries, tool-call fanout, context growth, and caching. Plus hard budgets you can actually enforce.