AI Briefing#prompt-caching
The latest AI and developer news about #prompt-caching, with the original source and a short summary.
The latest AI and developer news about #prompt-caching, with the original source and a short summary.
A comparative analysis of token usage and cache efficiency between Claude Code and OpenCode found that Claude Code's overhead was far higher.
Musinsa introduced Prompt Caching and cut LLM API costs by 64% while achieving a 98% cache hit rate.
This explains how Claude Code's prompt caching works and the cost differences depending on TTL settings.
Deep Agents automates prompt caching strategies that differ across model providers, reducing the operating costs of AI agents.
GitHub Copilot boosts the efficiency of agentic tasks through prompt caching and automatic model selection (Auto).
This covers how to run OpenAI models on Amazon Bedrock via the OpenAI-compatible API and leverage various advanced features.
Anthropic released Claude Opus 4.8, which offers improved honesty and code accuracy.
When using Claude Code, certain actions can invalidate the prompt cache, causing costs to spike by 12.5x.
It's advantageous to keep a Claude cache alive if you plan to write to it again within 62.5 minutes.
This analyzes the cost efficiency of using Claude Code subagents depending on whether token caching is applied.
By applying a caching technique that computes long prompts only once during RL training, training efficiency was significantly improved.
Opus 4.7's new tokenizer increased real-world costs by 12-27%.
For a single agent, the Batch API's latency is fatal, and cost efficiency only grows meaningful when multiple agents are bundled together.
Evidence has emerged that Anthropic lowered Claude Code's default prompt cache TTL from 1 hour to 5 minutes.
The OpenClaw tool for Anthropic's Claude models is supported again, supporting the latest features such as Claude 4.6's adaptive thinking and prompt caching.
A Korean-language compilation of Claude Code Harness analysis material has been shared as a web document.
Claude 4.7 produces 1.3–1.45x more tokens while showing a modest improvement in instruction compliance.
Claude 4.7 uses 1.3–1.45x more tokens, while strict instruction following improved slightly.
A case of running Qwen 3.6 35B on an RTX 2000 Ada 16GB at 400 t/s prompt and 24 t/s generation.
Cloudflare accelerated inference for massive LLMs through PD disaggregation, prompt caching, and KV-cache optimization.
The core of the Claude Code performance-degradation controversy is less about model nerfing and more about hidden operational variables.
Anthropic's Claude Code prompt cache TTL change and large-context costs have fueled complaints about faster quota depletion.
The shortened Claude Code cache TTL caused a sharp spike in cache regeneration costs and quota consumption.