Claude API Adds Inline Tools Without Breaking the Prompt Cache
Claude API inline tools can be added, replaced or pinned during a conversation without editing the original tools array or invalidating prompt caches.
Author: Abdessalam Alaoui
I'm Abdessalam Alaoui, founder of Musthave.AI. I write practical guides to AI models, tools, generative media, and ways real people make money with this stuff. No press-release fluff. 20+ years shipping digital products; you'll also find my work around HowTik, Yumasty, and IncomeHelpful.
JetBrains Air Expands From AI IDE to Multi-Vendor Agent Governance
JetBrains Air connects coding agents, team workflows, policy, auditability and cost controls. See what is available now and what remains a roadmap.
Grok 4.7 Jet Engine Demo Is Impressive, but It Is Not a Simulation
The Grok 4.7 jet engine visualizer is a polished interactive prototype. Here is what the viral clip proves, what remains unverified and how to…
Claude Opus 5.5 Targets Fable-Level Work at 40% Lower Cost
Claude Opus 5.5 lowers token and cache-read prices while targeting Fable-level work. See costs, benchmarks, access and a safe upgrade plan.
GPT-6 Sol and Luna Cut API Prices in Half
GPT-6 Sol Luna models bring lower API prices, improved caching and wider Codex access. Here is how builders should compare and route them.
IBM AI Invisible Work Study Finds Employees Are Validating and Correcting More
The IBM AI invisible work study finds validation, correction and exception handling are growing, while oversight skills and incentives lag.
AI Language Coalition Sets Five-Year Goal for 3.4 Billion People
The AI language coalition brings 60 organizations behind a five-year goal for voice and language access, but governance is still being designed.
BMW Cloud Cost Anomaly Detection Covers 14,000 Accounts for About $50 a Month
BMW cloud cost anomaly detection checks 14,000 accounts daily with Prophet, layered thresholds and owner feedback for about $50 in monthly compute.
Benchling AI Agent Security Blocks DNS Exfiltration From Generated Code
Benchling AI agent security isolates generated scientific code in a separate AWS account with DNS, storage and job-scoped credential controls.
Tempus Pays $12M to License Recursion’s RNA Foundation Model for Oncology
The Tempus Recursion RNA model agreement licenses TxFM for oncology for two years and rewrites a data deal with $42M in committed fees.
EU Data Center Energy Ratings Proposal Targets AI Power and Water Use
EU data center energy ratings would compare electricity and water use as AI demand grows, while minimum standards remain only a consultation.
Fastly AI Firewall Launches With a Streaming Inspection Gap
Fastly AI Firewall and Runtime Control add model routing, virtual keys and budgets, but streaming responses bypass output inspection.