Skip to main content
株式会社オブライト

Column

Useful articles about SEO, Web Development, and IT

771 articles

AI2026-10-09
Whistle Explained: Cactus's 16.9MB Speech-to-Text Model
Whistle is Cactus's 16.9MB open speech-to-text model for CPU, Apache 2.0. It supports 7 languages but not Japanese. Speed vs Whisper base, setup, best uses.
音声AIローカルAIオープンソース+2
Business DX2026-10-09
Migrating Access Databases: 5 Options and Costs for SMBs
Moving an Access customer or order database? Compare five options: keep Access, kintone, Power Apps, SaaS or a web rebuild, plus costs and a 5-step plan.
中小企業DXシステム発注開発費用+1
AI2026-10-08
Claude Haiku 5.5 Explained: $0.10/$0.50 Pricing, Benchmarks vs Haiku 4.5, and a Migration Guide (October 2026)
Claude Haiku 5.5 (Oct 7, 2026): $0.10/$0.50 per 1M tokens, 1M context, about a tenth of Haiku 4.5. Covers the 100K 5x price tier and 10 breaking changes.
AnthropicClaudeAPI料金+2
Business DX2026-10-08
Lost or Stolen Company PC or Phone: First-Hour Response and Costs
First-hour steps after a lost or stolen company PC or phone: police report, password reset, remote wipe, and breach reporting. Includes costs and checklists.
中小企業DXITリスク対策ひとり情シス+1
AI2026-10-07
Mistral Large 4 Requirements: VRAM & GPU for the 1T MoE (FP8 ~1TB)
Mistral Large 4 (1T MoE, 49B active) needs an estimated ~1TB at FP8, ~500-580GB at 4-bit. GPU, Mac and API cost compared. Updated Oct 2026; weights due soon.
Open Weight LLMMoERequirements+3
Business DX2026-10-07
Shadow IT and BYOD Policy for SMBs — Rules and Costs
Shadow IT is work done on tools or devices the company does not know about. Examples, risks, ban vs allow vs company phones, MDM costs, a policy checklist.
中小企業DXITリスク対策ひとり情シス+1
Software Development2026-10-07
What Is HyperFrames? HeyGen's Open-Source HTML-to-Video Framework for AI Agents (Setup, Usage, vs Remotion)
HyperFrames is HeyGen's Apache 2.0 framework that renders HTML into deterministic MP4. Oct 2026: setup, Claude Code skills, Remotion comparison.
開発ツールオープンソース動画+3
AI2026-10-06
Reflection Beam Requirements: VRAM & GPU for the 501B MoE (FP8 ~501GB, 8x H100+)
Reflection Beam (501B MoE, 23B active) needs an estimated ~501GB at FP8, ~250-290GB at 4-bit. GPU, Mac and CPU-offload fits as of October 2026; weights not out.
Open Weight LLMMoERequirements+4
Business DX2026-10-06
IT Onboarding Checklist for New Hires — SMB Guide and Costs
IT onboarding means getting a new hire's PC, accounts and permissions ready before day one. Timeline checklist, per-person costs, roles and common mistakes.
中小企業DXひとり情シスITリスク対策+1
Software Development2026-10-05
RemoveMacAI: Disable Apple Intelligence on macOS 27, Reclaim 12GB
RemoveMacAI is an MIT-licensed CLI that turns off Apple Intelligence on macOS 27 and deletes about 12GB of models. Install, commands, caveats, MDM comparison.
macOSApple Intelligence開発ツール+3
Business DX2026-10-05
Windows Server 2016 End of Support: SMB Options & Costs
Windows Server 2016 support ends Jan 12, 2027. SMB options: new server (800K-2M yen), NAS, Microsoft 365 or Azure VM. Costs, schedule, checklist and mistakes.
中小企業DXひとり情シスITリスク対策+2
AI2026-10-05
What Is Strata? Run Qwen3.8 Flash Next (125B) on an RTX 4090 — Requirements and Setup
Strata is an MIT-licensed engine running Qwen3.8-Flash-Next (125B MoE) on a 12GB gaming PC: 94 tokens/s on an RTX 5070. Oct 2026 specs and setup.
ローカルLLMVRAM開発ツール+3
1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465