Merge Fusion says one prompt and a judge model can beat frontier systems for 1/4 the cost
thisguyknowsai · x · 2026-07-24
A new product called Merge Fusion claims it can beat frontier models at a quarter of the cost by using a panel of models plus a judge model.
- One prompt is sent to multiple models.
- A judge then synthesizes a single answer that outperforms any individual model.
- The launch claims Fusion beats Fable 5 outright, costs 1/4 as much, and sets the top score on the DRACO benchmark.
More from Venture
- Anthropic S-1: 11,900+ customers spend $100k+ yearly, ~250 spend $10M+ — RebeccaBellan · 2026-10-27
- Anthropic's S-1 leads with AI curing diseases claim, doubling down despite pushback — nlarusstone · 2026-10-27
- Anthropic's S-1 admits open-weight models will squeeze its pricing, rivals gloat — teortaxesTex · 2026-10-27
- PitchBook: vertical AI apps yield ~65% gross margins, better ROI than horizontal — MartinGTobias · 2026-10-27
- He gave his AI agent a phone number and email — then hired it as social media manager — heyshrutimishra · 2026-10-27
- Firecrawl launches Alexandria company-data API and announces $75M Series B — devdigest · 2026-10-27