The cheapest credible frontier-adjacent option — the price floor of the whole market.
V4-Flash is the lightweight sibling of V4-Pro and sets the price floor for the whole market: $0.14 / $0.28 per million tokens, MIT-licensed and fully self-hostable. Its 284B/13B-active mixture-of-experts is far easier to run than V4-Pro while landing near it on many coding tasks, and it keeps the 1M-token context.
There is no standalone independent intelligence score for Flash in this survey, and on the hardest reasoning it sits below V4-Pro; the hosted API shares the same peak-hours surcharge, and it is text-only. Those are unsurprising limits at this price.
For cost-sensitive, high-volume, or coding-heavy workloads — especially where self-hosting a smaller open model is attractive — V4-Flash is one of the most economical credible options available, with V4-Pro the step up when a task needs the extra reasoning.
FIELD NOTE — Lighter 284B/13B sibling of V4-Pro; near it on many coding tasks.
VISIT OFFICIAL SITE →