Latest cybersecurity dispatches

Fresh reporting on active vulnerabilities, security patches, incident response, and threat research for defenders.

Articles: 407
Page: 50 / 51
Desks: 9

Data graphic: published file or download sizes of open-weight model builds against a 24GB Apple Silicon Mac, led by the practical pick Qwen3.8-27B in 4-bit MLX at 16.1GB, with Meta Muse Glimmer 30B NVFP4 at 17GB, Gemma 4 26B A4B at 14.2GB, Gemma 4 12B Unified at 6.98GB, Gemma 4 E4B at 4.22GB, and Laguna XS 2.1 at 20.3GB flagged because Poolside specifies a 36GB Mac. It also notes a 16K–32K working context for 27B–30B models, NVFP4 then 4-bit MLX then Q4_K_M as the best quantizations, and that Devstral Small 2 was retired on 31 March 2026.

Best Open-Weight AI Models for a 24GB Apple Silicon Mac in 2026

Qwen3.8-27B at 4-bit MLX and Meta Muse Glimmer 30B are the picks for a 24GB Apple Silicon Mac. Devstral Small 2 was retired in March 2026.

May 17 · 21 min

Data graphic: Claude Code vs Codex limits. ChatGPT Pro $200 drops from 20x to 10x Plus usage from 30 Oct, while Claude Max 20x stays at 20x Claude Pro per session. Bars show published Codex GPT-6 Astra limits of 25–225 local messages per 5 hours on ChatGPT Pro $100 and 100–900 on Pro $200 until 29 Oct, next to tier cards for Claude Max 5x and 20x $100 and $200 and ChatGPT Pro $100 and $200.

watchAI Security AI Coding Tools

Claude Code vs OpenAI Codex Limits: Which $100 or $200 Plan Is Better for Developers? [UPDATED]

Codex still publishes clearer per-model limits and Claude still reasons deeper, but OpenAI's $200 Pro tier drops from 20x to 10x Plus usage on 30 Oct.

May 17 · 17 min

Claude Max 100 versus 200 Claude Code plan usage comparison for developers

watchAI Security AI Coding Tools

Claude Max $100 vs $200: Which Plan Gives Developers More Real-World Claude Code Usage?

Claude Max 5x is the better-value plan for many solo developers, while Max 20x gives heavier Claude Code users more room for long sessions, large repositories and agentic workflows.

May 17 · 16 min

ChatGPT Pro 100 versus 200 Codex plan usage comparison for developers

watchAI Security AI Coding Tools

ChatGPT Pro $100 vs $200: Which OpenAI Codex Plan Gives Developers More Real-World Usage? [UPDATED]

OpenAI's $100 Pro plan is the best Codex value for solo developers. Pro $200 is open again but drops from 20x to 10x Plus usage on 30 Oct 2026.

May 17 · 10 min

Fragnesia CVE-2026-46300 Linux kernel page cache local privilege escalation path

highExploits Linux Kernel

Fragnesia Linux Kernel Flaw Gives Local Attackers a Public Path to Root

Fragnesia, tracked as CVE-2026-46300, is a Linux kernel local privilege escalation flaw involving ESP/IPsec page-cache corruption, with public proof-of-concept code increasing patch urgency.

May 17 · 6 min

React Server Components CVE-2026-23870 denial-of-service path affecting Next.js App Router

highExploits Web Framework Security

CVE-2026-23870: React Server Components DoS Flaw Hits Next.js App Router Apps; How to Reproduce Safely and Patch

A high-severity React Server Components denial-of-service flaw can let unauthenticated attackers degrade vulnerable Next.js App Router apps through crafted Server Function requests.

May 17 · 6 min

LiteLLM CVE-2026-42208 AI gateway SQL injection incident map

criticalAI Security AI Gateway Security

LiteLLM CVE-2026-42208: Critical Pre-Auth SQL Injection Exposes AI Gateway Secrets

Critical LiteLLM CVE-2026-42208 pre-auth SQL injection exposes AI gateway databases, virtual keys and upstream model-provider credentials.

May 16 · 6 min

Data graphic: user-reported Ollama Cloud speeds of 22 tokens per second for MiniMax 2.7, 14 for Qwen 3.5, 11 for GLM 5 and 8 for GLM 5.1, with the slowest rate, 8 tok/s, as the headline number, 2.75x slower than the fastest. It also shows how long a 4,000-token reply takes to stream at each speed, from 3m 02s up to 8m 20s, and that the Free, Pro and Max plans run 1, 3 and 10 cloud models at once, with extra requests queued.

watchAI Security AI Cloud

Ollama Cloud Slowdown Frustrates Users as Low Token Rates Raise Questions About Hosted Open-Model Economics

User-reported token rates as low as 8 tokens per second show why Ollama Cloud needs clearer performance metrics, queue visibility and paid-tier expectations.

May 15 · 8 min