Ripple-65B: 65B dense model claims open-weight agentic coding SOTA, beats 500B+ MoEs
simonguozirui · x · 2026-11-05
Ricursive AI releases Ripple-65B Preview, a from-scratch pretrained 65B dense LLM claiming new state-of-the-art among US open-weight models on agentic coding tasks, beating far larger models like Muse Spark Open-816B and Inkling-975B. Training engineer @ethanboneh shares nonobvious lessons: dense models can beat larger MoEs downstream even at matched perplexity (attention density, less memorization); mixed RL becomes extremely efficient once infra and synthetic environment pipelines work; and swarms of small agentic models can turn qualitative dataset questions into programmable measurements, creating a data-quality feedback loop.
More from Models
- Ripple claims frontier-level open-weight model trained in 32 minutes at one-tenth cost — annadgoldie · 2026-11-05
- Developers report Grok API instability: 13.58% error rate and 13.6s p90 latency in dashboards — ChrisUniverse · 2026-11-05
- Ramp's post-trained Qwen beats Opus on finance QA; 10,000-agent swarm finds 7 critical bugs — AI Engineer · 2026-11-05
- Vals AI: four of seven frontier models keep 80%+ scores at lowest reasoning effort for half the cost — ZacharyHuang12 · 2026-11-05
- Sailing sim race shows Chinese models lag frontier: "cheap tokens ≠ cheap agents" — shekitup · 2026-11-05
- DIY logic-puzzle benchmark: Qwen 30B-A3B beats Gemma A4B at reasoning — Robert__Sinclair · 2026-11-05