Open-weight models hit 78.4% of token volume on Vercel AI Gateway
GaryMarcus · x · 2026-09-19
Vercel CEO Guillermo Rauch shared AI Gateway stats: 78.4% of that day's token volume came from open-weight models, vs 21.6% closed.
Spend usually tells a different story, but Moonshot AI and DeepSeek ranked #3 and #4 by spend; combined with Z.ai, their total surpasses OpenAI (#2). He notes this is inference spend across providers (mostly US), not revenue going directly to the open-weight labs. Gary Marcus quote-posted it to mock Dario Amodei's reported $2T IPO plans.
Related event: Open-source models hit 78% of tokens on Vercel gateway(4 posts)→
More from Models
- Fruit fly connectome chess model beats Jev 4-1 in 10 games, with a playable demo site — maximelabonne · 2026-09-20
- Bonsai 2 27B safety guardrails reportedly cut SWE-bench and Terminal-bench scores by ~20 points — julianharris · 2026-09-20
- MiMo-V2.6 livestreams its RL run at ~2B tokens/step as Stanford's Marin pretrains in public — stanfordnlp · 2026-09-20
- Jev as a New Primitive: Cheaper Scaling, Better Verifiers and Agent Harness Experiences — omarsar0 · 2026-09-20
- What justifies paying more for an AI model? The 16-second break-even math — zeuslac · 2026-09-20
- Skeptical deep dive confirms Humanity's Last Exam errors; official o3-mini grader marked right answers wrong every time — paul_cal · 2026-09-20