[태그:] Developer Tools
Perplexity Let AI Write Its Own Search — And It Won
Perplexity’s Search as Code lets AI write its own Python search pipelines in a sandbox — beating OpenAI and Anthropic while cutting token costs up to 85%.
OpenAI’s Code Agent Just Moved Into Amazon’s House
GPT-5.5 and Codex are now GA on Amazon Bedrock with pay-per-token pricing and no seat licenses. See what the end of Azure exclusivity changes for you.
Anthropic Taught Claude to Dream — Tasks Got 6× Faster
Anthropic shipped dreaming for Claude agents — Harvey’s task completion jumped 6× and Wisedocs cut review time 50%. Here’s what changed.
AI Just Found 10,000 Bugs Nobody Can Patch Yet
Mythos Preview found 10,000+ critical vulnerabilities in weeks. Cloudflare alone surfaced 2,000—Mozilla 271. Here’s why patching can’t keep up.
12 Million Tokens for $8 — Yes, Really
SubQ just shipped a 12-million-token AI model that runs for $8 — versus $2,600 on frontier models. Find out who built it and why labs are quiet.
OpenAI Slipped GPT-5 Brains Into Your Microphone
OpenAI launched 3 new voice models with GPT-5-class reasoning and a 128K context window. Here’s why Zillow’s call success jumped 26 points.
OpenAI Just Made Every Other AI Look Outdated
GPT-5.5 ships with real computer-use and multi-step agent workflows at $5 per million tokens. OpenAI’s fastest model cadence yet signals the agentic AI era has arrived.
This $0.14 Model Almost Matches GPT-5.4
DeepSeek V4-Pro delivers 80.6% on SWE-bench at $1.74/M tokens, within 0.2 points of Claude Opus 4.6. The 1.6T-parameter open-weights model and its $0.14 Flash sibling are reshaping AI economics.
Snap’s 65% AI Code Claim Just Fell Apart
Snap fired 1,000 workers citing AI writes 65% of code. But the cuts hit product managers, not engineers. Activist investor pressure and $500M savings tell a different story.
OpenAI Doubled the Price — Developers Don’t Care
GPT-5.5 scores 82.7% on Terminal-Bench 2.0 and 90.1% on BrowseComp at double the price of GPT-5.4. Find out why developers are paying anyway.
