Llm Kosten

OpenAI has released the first benchmark results for "Jalapeño," its custom inference ASIC built with Broadcom. The chip delivers up to 1.9x more work per watt and significantly lower latency, especially for interactive and agentic workloads.

Claude Fable 5.1 maintains the same pricing as its predecessor but introduces a 75% drop in cache read costs and significantly lowers token usage per task. Here is a deep dive into what this means for agentic workflows and a copyable cost-per-task audit for your own stack.