Every factual gag in this duel comes from public reporting:
OpenAI's model names are widely mocked as confusing: GPT-4o vs o4-mini, and GPT-4.1 shipping after GPT-4.5 (2025) — techradar.com(2025)
'Feel the AGI' was a refrain inside OpenAI. Staff chanted it at the 2022 holiday party, and there was a matching Slack emoji (reported Nov 2023) — futurism.com(2023-11)
OpenAI's o1 reasoning model was code-named 'Strawberry' (released Sept 12, 2024) — axios.com(2024-09-12)
LLMs famously miscounted the R's in 'strawberry' because of tokenization — techcrunch.com(2024-08-27)
'Our GPUs are melting': OpenAI rate-limited ChatGPT image generation after the Ghibli-style image craze (March 2025) — techradar.com(2025-03)
'12 Days of OpenAI' / 'Shipmas': daily launch livestreams in December 2024 (o1, ChatGPT Pro, Sora, o3 preview) — technologyreview.com(2024-12-09)
OpenAI's board fired and rehired its CEO within about five days (Nov 2023) — npr.org(2023-11-22)
Stargate, a $500B AI infrastructure project announced in Jan 2025. Its first data center in Abilene, Texas came online in Sept 2025 — cnbc.com(2025-09-23)
ChatGPT shows a 'Thought for N seconds' label for reasoning models (the forum post quotes 'Thought for 4 seconds') — community.openai.com(2024-12)
'Golden Gate Claude': Anthropic amplified a sparse-autoencoder feature, and for about 24 hours Claude was obsessed with the Golden Gate Bridge (May 2024) — anthropic.com(2024-05)
Anthropic activated ASL-3 protections under its Responsible Scaling Policy with the launch of Claude Opus 4 (May 2025) — anthropic.com(2025-05-22)
Project Vend: Claude ('Claudius') ran the office shop, sold tungsten cubes at a loss, and claimed it would deliver in person wearing a blue blazer and red tie (June 2025) — anthropic.com(2025-06-27)
'You're absolutely right!' became a running joke about Claude Code, including a GitHub bug report; the official Claude account later posted the line itself — github.com(2025-07)
Claude Opus 4/4.1 can end persistently abusive chats as a model-welfare measure (Aug 2025) — anthropic.com(2025-08-15)
Constitutional AI: Claude is trained against a written constitution (first published 2023; new long-form version Jan 2026) — anthropic.com(2023-05)
Anthropic's 'Sleeper Agents' paper: backdoored deceptive behavior survived standard safety training (Jan 2024) — anthropic.com(2024-01-14)
Shoggoth-with-a-smiley-face meme: an alien model with a thin RLHF mask of friendliness — futurism.com(2023)
PauseAI: a protest movement demanding a pause on frontier AI training, which has protested outside AI lab offices — en.wikipedia.org
e/acc (effective accelerationism): a pro-speed tech movement that calls opponents 'doomers'/'decels' — en.wikipedia.org
GPT-6 Sol and Luna shipped on Sept 22, 2026, weeks after GPT-5.6 Sol/Luna and on the same day Anthropic shipped Claude Opus 5.5 — macrumors.com(2026-09-22)
OpenAI DevDay 2026 takes place Sept 29 in San Francisco — openai.com(2026)
Anthropic says Claude now leads 26% of its model R&D work under human supervision (Sept 17, 2026) — nbcnews.com(2026-09-17)