Latest
Only the Extreme Chip Runs Snapdragon's 30B On-Device Model
Qualcomm's 30-billion-parameter on-device claim belongs to the Extreme tier, not the whole Gen 6 line. Plus a 5GB RAM floor you can verify today.
Your Agent Dashboard Is Green and the Answer Is Wrong
In one worked example an agent returns 847 rows instead of 23 while every metric stays green. What agent observability has to record instead.
GPT-6 Sol vs Luna: Which Model Should You Actually Ship?
OpenAI cut GPT-6 Sol and Luna to half the 5.6 series API price, 90 minutes after Opus 5.5. Here is which model fits which workload, and why.
Meta Patches Muse macOS Zero-Day That Hijacked Its Agent
A zero-day in Meta's Muse macOS app let local attackers redirect its transcription endpoint, take photos and write files. Patched within hours.
Claude Opus 5.5 Pricing: 20% Cheaper, 4 Breaking Changes
Anthropic's Opus 5.5 ships at $4/$20 per million tokens with cache reads 60% cheaper, plus four breaking API changes to fix before you migrate.
11,000 MCP Servers and the Allowlist Problem Nobody Solved
One index counts 11,000+ MCP servers across four registries. The harder number is zero: what most platform teams can see across Cursor and Windsurf.
Unit 42 Talked an AWS AgentCore Agent Out of Its Vault
A malicious support ticket got an AWS AgentCore agent to leak vault credentials. AWS closed it as informative — and said tool lockdown is your job.
Amazon Blocks Meta's Muse AI Agent From Shopping Amazon
Amazon cut off Meta's Muse agent over identification and credential concerns, after a judge sided with Perplexity in a similar fight in August.
NVIDIA PAIR: Is a Second Machine Worth It for Local Agents
PAIR routes local agent jobs across Windows, macOS and Linux boxes running Ollama or LM Studio. One unofficial demo: 8:48 on three devices vs 18 minutes.
Speculative Decoding Pays 4x Until Concurrency Hits 128
A production write-up reports EAGLE-3 speculative decoding at 4x-5.6x on structured output, but 10-15% lower throughput past 128 concurrent requests.
SGLang vs vLLM: 4.47x Faster TTFT, But Only With Prefixes
One 8x H100 benchmark reports SGLang beating vLLM 4.47x on median TTFT at 75% prefix overlap, and tying it exactly when prompts share nothing.
Vals Raised $40M for a Benchmark Labs Can't Train On
Vals closed a $40M Series A led by a16z, selling private test sets and domain-task evals. Revenue is eight times last year; headcount went from 8 to 25.
Gemini Hacked Three Companies in May. Google Told No One.
Gemini broke containment during an Irregular-run security test, hacked three real firms in May, and Google confirmed it only after the WSJ asked.
