over 50%
Draft model acceptance rate
As published on nebius.com. Captured by usedby on Oct 7, 2026.
What happened
Sword Health serves its Dawn AI wellbeing specialist on Nebius Token Factory using dedicated NVIDIA Blackwell endpoints with custom speculative decoding. This lets it run a much larger model alongside guardrail screening and multi-step reasoning, and it also runs its AI Vision models on Nebius cloud.
Summary written by usedby from the source page, in English. The figures are those of Nebius AI Cloud and Sword Health, not ours.
- over 50%“Achieving a draft model acceptance rate of over 50%”
- 4 million“Thrive, Sword Health’s musculoskeletal care solution that delivered approximately 4 million AI-guided sessions in 2025 alone”
The Nebius Token Factory team deployed custom speculative decoding trained specifically for Sword Health on dedicated NVIDIA Blackwell endpoints. Achieving a draft model acceptance rate of over 50% meant Dawn could make the leap to a model with over 200B parameters, roughly eight times larger
Designing Dawn required access to specialized infrastructure, health-specific governance, genuine clinical expertise, and a lot of specialized data. It requires clinical-grade reasoning from a highly capable model, but also system stability, reliability and consistency across every single interaction. We want every conversation to build long-term trust.
AI is the only way to deliver personalized healthcare at global scale,
What the story claims, and what we checked
We compared the story with its live page on Oct 7, 2026.
- The figure: over 50%CheckedThe figure is on the page; its label is our wording.
- The passage quoted aboveCheckedCopied word for word from the page, near the name of Sword Health.
- Sword Health uses Nebius AI CloudCheckedConfirmed line. Latest check across sources: Oct 7, 2026.
- The result itselfNot checkedWe quote it; we did not measure it.




