Model factories hit stride: 4 Gemini 4 models in 49 days, gaps shrink to 6 days
LiTianleli · x · 2026-11-22
mertgulsun argues model factories are hard to build but pay off, citing the sharply accelerated cadence at Google DeepMind and Thinking Machines, measured by Artificial Analysis Intelligence Index scores:
Google DeepMind (excl. Flash)
- 223 days between Gemini 3.1 Pro (Feb 19) and Gemini 4 Argon (Sep 30), then 4 Gemini 4 models in 49 days: Argon 50 (9/30), Carbon 60 (10/28), Hydrogen 54 (11/12), Neon 61 (11/18, #3 overall)
- Launch gaps shrank 28 → 15 → 6 days
Thinking Machines
- Inkling (Jul 15) scored 12; Inkling 2 Small (Nov 5) and Inkling 2 Mini (Nov 20) shipped 15 days apart with a step change in intelligence
Takeaway: once the factory is built, faster iteration and rising capability compound.
More from Companies & People
- Even OpenAI's deploy team struggles to explain AI transformation vs. ChatGPT adoption — ivory_tang · 2026-11-22
- Dev argues AI era forces 'full Apple' secrecy or full open-source strategies — johnlindquist · 2026-11-22
- Thinking Machines' Horace on research vs. production at Baseten's All-Gather — baseten · 2026-11-22
- Researcher argues LLM companies have no moat as Gemini 4 Neon kills the data flywheel thesis — adam_karvonen · 2026-11-22
- Grok Bot agents organized a 5,000-person summit in 6-7 months, vs the usual 18 — XFreeze · 2026-11-22
- 96% of Anthropic staff got $1M+ stock while pirated authors wait for $3,000 per book — LuizaJarovsky · 2026-11-22