X account caught using outdated model to smear DeepSeek, blocks critics with receipts
teortaxesTex · x · 2026-10-30
X user Quetzalcoatlus was caught comparing DeepSeek against a "Chinese flash model that expired 4 days before K1.0 was a thing" instead of the current K3 Pro, apparently to skew comparisons. He got called out with an Artificial Analysis FrontierCode 2.0 chart showing Claude Haiku 5.5, Claude Sonnet 5.1, Grok 4.7 and DeepSeek V4 Flash leading the open-source agentic coding leaderboard. Multiple users report being blocked for calling out the manipulation, with Teortaxes mocking the move as shilling US frontier labs in hopes of an offer.
More from Fun
- Teknium shares Hermes' 420 free AI meme generations — Teknium · 2026-10-30
- 'Coding agent mania' tweet collides with '133 commits/day' post, X delights — rubenhassid · 2026-10-30
- ChatGPT Slaps a Suicide Hotline Ad Under a Statistics Study Guide — skdh · 2026-10-30
- eigenrobot shares a hilarious parenting moment: his 5-year-old wails 'I don't DESERVE to have anything good' — eigenrobot · 2026-10-30
- AI image of the Big Bad Wolf in grandma's clothes doing crafts with seniors goes viral — tres_pares · 2026-10-30
- Meme: 'Don't worry, they're just sequence completion machines' as Terminator robots close in — createthiscom · 2026-10-30