<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:news="http://www.google.com/schemas/sitemap-news/0.9">
<url><loc>https://news.contentslab.ai/2026/10/25/llamaindex-ocria-1/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-25T18:20:00+09:00</news:publication_date><news:title>PDF 속 표 · 차트를 그대로 옮기는 4B 모델 — LlamaIndex 'OCRia-1', 한 쪽 2.2초</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/vals-inference-provider-index/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T11:10:00+09:00</news:publication_date><news:title>같은 오픈 모델도 어디서 돌리냐에 따라 비용 최대 2.3배 — Vals 'Inference Provider Index'</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/epoch-benchmark-reviews/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T08:05:00+09:00</news:publication_date><news:title>Epoch AI 벤치마크 점검표 — 22개 중 12개가 'Flawed', SWE-bench Verified · HLE · OSWorld도</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/openai-offline-training-containment/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T05:25:00+09:00</news:publication_date><news:title>OpenAI, 최신 모델 학습에서 실시간 인터넷을 끊었다 — 9월 20일 사고 뒤 '오프라인 세상'에서 학습</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/chroma-joins-replit/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T05:25:00+09:00</news:publication_date><news:title>벡터 DB Chroma, Replit에 합류 — Chroma Cloud는 11월 22일 쓰기 · 12월 22일 읽기 중단</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/claude-code-mcp-tasks/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T04:20:00+09:00</news:publication_date><news:title>Claude Code, 오래 걸리는 MCP 작업을 서버에 맡기고 계속 일한다 — 'MCP tasks' 공개 베타</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/tencent-hy-nomas/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T03:15:00+09:00</news:publication_date><news:title>말로 이어 가는 영상 속 세계 — 텐센트 'Hy-Nomas' 오픈소스 공개, RTX 4090 한 장 설정도</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/kimi-k3-1-pro/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T03:15:00+09:00</news:publication_date><news:title>코딩 · 3D · 금융에 강해진 3T급 오픈 모델 — Moonshot 'Kimi K3.1 Pro' 공개</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/ai-sdk-web-fetch-tool/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T03:15:00+09:00</news:publication_date><news:title>Vercel AI SDK에 공식 '웹 페이지 읽기' 도구 — 어느 모델이든 주소만 주면 마크다운으로</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/devin-desktop-agent-window/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T02:25:00+09:00</news:publication_date><news:title>Devin Desktop 새 '에이전트 창' — 작업 세션 최대 8개를 한 화면에 나란히</news:title></news:news></url>
<url><loc>https://news.contentslab.ai/2026/10/24/nvidia-nemotron-3-5-bolt/</loc><news:news><news:publication><news:name>CONTENTSLAB NEWS</news:name><news:language>ko</news:language></news:publication><news:publication_date>2026-10-24T02:20:00+09:00</news:publication_date><news:title>약 4GB로 도는 대화형 에이전트 모델 — NVIDIA 'Nemotron 3.5 Bolt E4B 4GB' 공개</news:title></news:news></url>
</urlset>
