I've seen some confusion online on how to run llama.cpp with MTP (Multi-token prediction) in the simplest way possible. ICYMI, MTP is a new flavor of speculative decoding built-in to the model itself, that ~2x your tokens per sec for most use cases. 2x generation speed = Truly
TOOLS
-
Cloudflare evaluates Anthropic’s Mythos model on internal code repositories
By
–
Cloudflare pointed Anthropic's Mythos Preview at 50+ of their own repos. They call it a step-function forward "Mythos Preview is a real step forward, and it's worth saying that plainly before getting into anything else." The big finding isn't the bugs it caught – It's that the
-
Service pools free tiers of dozens of AI models into one API
By
–
800 million tokens per month. Free. Someone just combined the free tiers of dozens of AI models into a single API. You input your API keys and it automatically manages: • routing between models
• rate limits
• fallback when a provider hits the limit
• load balancing -
Using Claude Code to audit personal devices
By
–
A nice way to stay safe is to ask Claude Code to audit your devices I do same on my VPS servers, so today I tried it on my MacBook Pro and it's pretty good at it too It founds lots of stuff that was not secured, I actually forgot to enable FileVault when I got this new MBP in
-
ClickUp AI Brain Builds Interactive Apps from Prompts
By
–
ClickUp's AI Brain will soon be able to build fully interactive apps from a single prompt, with live data integrations. Some examples > Sprint tracker – to reshape your team operations into an analytics dashboard for stakeholders. > CRM tool – to have a custom dashboard
-
Anthropic Announces Self-Hosted Sandboxes for Claude Agents
By
–
Anthropic announced self-hosted sendboxes and MCP tunnels for Claude Managed Agents during its "Code with Claude" event in London.
— 🚨 AI News | TestingCatalog (@testingcatalog) 19 mai 2026
> With self-hosted sandboxes, you keep sensitive files, packages, and services in your own infrastructure or with a managed sandbox provider.
>… https://t.co/Nap7QEosHO pic.twitter.com/AeSO0QMOpTAnthropic announced self-hosted sendboxes and MCP tunnels for Claude Managed Agents during its "Code with Claude" event in London. > With self-hosted sandboxes, you keep sensitive files, packages, and services in your own infrastructure or with a managed sandbox provider. >
-
PolyAI Opens Platform for Rapid AI Agent Deployment
By
–
PolyAI opened their platform . Two months free, ten minutes from sign up to a deployed agent. The infrastructure is the same one running calls for Marriott, FedEx, and most of Vegas.
-
New ADK released for Claude Code integration
By
–
There's a company you've almost certainly never heard of that's quietly handled a billion conversations across 75 languages.
— AI Highlight (@AIHighlight) 19 mai 2026
You've probably talked to their AI without knowing it.
Today they shipped an ADK that plugs into Claude Code and nobody is talking about it. pic.twitter.com/YEessO6THqThere's a company you've almost certainly never heard of that's quietly handled a billion conversations across 75 languages. You've probably talked to their AI without knowing it. Today they shipped an ADK that plugs into Claude Code and nobody is talking about it.
-
Regression reported in Anthropic’s Opus 4.7 model performance
By
–
Opus 4.7 is ignoring project instructions, skipping MCP servers, and burning through usage limits on tasks Opus 4.6 handles first try. Dozens of confirmed reports. Anthropic acknowledged the regression. The failures are specific:
→ Project instructions ignored entirely
→ MCP -
Using cross-tool AI workflows for coding efficiency
By
–
Using Codex to make Claude better is exactly the kind of cross-tool workflow that's becoming the actual edge.