Open-weight models take under 1% of tokens on T3 Code, debunking margin-erosion FUD
thedealdirector · x · 2026-11-23
The author debunks 2026's claim that open-weight models erode frontier margins: OpenAI and Anthropic run at $100B+ run rates, and on T3 Code only 6% of users tried an open-weight model in the last 7 days (down from 16%), with under 1% of token share. Full-size open models underperform at near-frontier API prices; quantized versions have poor value, need $5,000+ hardware, and self-hosting wastes productive hours versus a $200 frontier subscription. The only real case is air-gapped enterprise/government — better served by BYOC, e.g. Anthropic running custom Opus on your own hardware.
More from Models
- Open vs closed models: the six user-side advantages of self-hosting, summed up in a viral tweet — gnukeith · 2026-11-23
- OpenAI Publishes Internal Pretraining Perplexity Data, Calls for Industry-Wide Safety Limits — sama · 2026-11-23
- Same weights, one V100: inference engines differ up to 5x in tokens/s — PuzzleheadedArm3732 · 2026-11-23
- Tesla FSD V15 clip raises questions: how far can pure vision really go? — ezshine · 2026-11-23
- Do Emotions Drive Agent Behavior? Speculation That Blocking Strong Feelings Constrains Agency — mimi10v3 · 2026-11-23
- Writers Push Google for Gemini 18+ Age Verification and Content Toggles to Stop Exodus to Grok — Hikkynure · 2026-11-23