OpenAI Mocked Again for Cramped, Unreadable Benchmark Charts
lyraxana · x · 2026-11-18
- Users are mocking OpenAI for releasing yet another poorly designed chart: an Agents' Last Exam v1.1 cost-vs-score plot crams all models into the $0–$15 range while the x-axis extends to $60.
- The chart lists models like GPT-6.2 Astra/Sol, GPT-6/6.1, Claude Fable 5.2, Claude Opus 5.5, and Gemini 4 Carbon (apparently a parody/satire screenshot, unverified).
- The wasted whitespace and useless scale make actual differences nearly unreadable, making it a running joke in the community.
More from Fun
- One Gemini App session yields a WebGPU simulation plus a scored 15s video — AI_Andrew · 2026-11-18
- Kalshi says Jensen Huang returned to X after Nvidia ouster, adding ~$100B — nima_owji · 2026-11-18
- Me Throughout All My Trading Losses — mfckr_eth · 2026-11-18
- Berkeley's Ken Goldberg: From 1994's crowdsourced Telegarden to agent-written robot code — lukas_m_ziegler · 2026-11-18
- Dev recreates Notre-Dame's nave in UE5 with GPT-6.2 Astra, organ track included — Dimillian · 2026-11-18
- MIRI President's Partner Runs Bay Area Kink Parties, Report Says — thedealdirector · 2026-11-18