Ripple ladder shows data beats architecture: 4x gains from new dataset vs 3x from new arch
finbarrtimbers · x · 2026-11-05
Finbarr Timbers highlights a striking finding from Ricursive Intelligence's Ripple model ladder: data matters more than architecture.
Ripple 1.6 vs Olmo 3 kept the same dataset but introduced a new architecture, yielding 3x improvement, while final Ripple vs 1.6 kept the same architecture but swapped in a new dataset, yielding 4x. Ripple-65B Preview, pretrained on 12T tokens with 1M context, scores 80.52 on TerminalBench 2.1 — US open-weight SOTA, beating Reflection Beam 501B (80.1) and Meta Muse Spark Open 816B (76.4) — reportedly at one-tenth the usual frontier training cost.
Related event: Ripple Ladder Experiments Suggest Data Matters More Than Architecture(2 posts)→
More from Infra
- Decentralized RL Works: Untrusted Miners Lift LiveCodeBench 39.5→58.7 on Bittensor Subnet — bittingthembits · 2026-11-05
- Atomic Chat launches open-source Local AI Server tying models and agents together — SucceededMind · 2026-11-05
- Developers report Grok API instability: 13.58% error rate and 13.6s p90 latency in dashboards — ChrisUniverse · 2026-11-05
- Even cutting forecasted data center growth by 60% leaves Nvidia supply constrained — robleclerc · 2026-11-05
- Free live paper discussion: why speculative decoding is fast despite doing more compute — doesdatmaksense · 2026-11-05
- Airbnb: fine-tuned 2B model matches frontier classifiers at 1% of the cost — Ahmad_Al_Dahle · 2026-11-05