Skip to content
MustHave.ai
MustHave Hoodie Seal
THE MUSTHAVE NEWSROOM

AI news

What’s new. What matters. What do you think?

Top stories

By member votes

Eleven v4 API Guide: Choose the Right Model, Endpoint and Cost

Eleven v4 and v4 Turbo are available through different API paths. The right choice depends on dialogue quality, latency, request size and credit use.

Gemini 4 Argon Is Now Available: API Access, Pricing and Output Limits

Gemini 4 Argon API access, pricing and output limits, plus a downloadable 18-check coding fixture pack with an explicit testing method.

Gemini Skills Replace Gems: Migration Dates, Limits and What to Save

Google is replacing Gemini Gems with Skills on different schedules for personal, business and education accounts. Here is what moves, what does not and what to back up.

Cloudflare Threat Signals Goes Free: RSS Limits and Human Review

Cloudflare Threat Signals is generally available with one free RSS feed and up to 30 days of data. Extracted indicators still need review before enforcement.

GKE Agent Sandbox RL SDK: What Google’s 45× Startup Claim Measures

Google’s sandbox SDK targets the waiting time in agentic reinforcement learning. Its 45× headline concerns tail startup latency under stated conditions.

Grok Team Bots: Shared Skills, Private Chats and Team Permissions

Grok Team Bots adds a shared operating model in public beta. Private conversations, team memory and shared connector credentials have different boundaries.

Cloudflare cf CLI: What Changes for Wrangler Users and AI Agents

Cloudflare’s cf CLI is in open beta. Start with account commands, understand the migration boundary and keep a tested route back to Wrangler.

Hugging Face Serge: Why Bug-Fixing Agents Need Verification

Hugging Face reports 29 fixes from its Serge workflow. The useful lesson is how a failure becomes a verified patch, a reviewed PR and an accepted fix.

ChatGPT Plugin Extensions: Which Surface Should Your App Use?

ChatGPT plugin extensions add sidebar apps, conversation panels, file handlers and desktop composer mentions. Here is a builder's map of each surface and its rollout limits.

OpenAI Dots: What the Always-On Agents Can Read, Do and Ask You to Approve

OpenAI dots can research connected apps in the background and carry out approved work. This guide separates read-only activity, actions, account access and rollout limits.

GPT-6.1 Sol: API Pricing, Long-Context Costs and a Migration Test

GPT-6.1 Sol is in the API, ChatGPT Work and Codex, but not Chat. Here is what its token rates mean for cached and long-context jobs, plus a migration test.