Ripple-65B Preview, a 65-billion-parameter open-weight model, scored 80.52 on TerminalBench 2.1 — beating models up to 15 times its size, including one with 975 billion parameters.
Ripple-65B Preview, a 65-billion-parameter open-weight model, scored 80.52 on TerminalBench 2.1 — beating models up to 15 times its size, including one with 975 billion parameters.
0 vue
Publié le 05/11/2026 20:19
0:29
Ripple-65B Preview, a 65-billion-parameter open-weight model, scored 80.52 on TerminalBench 2.1 — beating models up to 15 times its size, including one with 975 billion parameters.
1 commentaire
?
It topped Reflection Beam (501B params, 80.1), Meta's Muse Spark Open (816B, 76.4), Thinking Machines' Inkling (975B, 63.8), and NVIDIA's Nemotron 3 Ultra (550B, 56.4).
Ricursive Intelligence says it built and trained Ripple-65B for $32 million total — about a tenth of what it estimates Reflection Beam cost. Reflection's team has reportedly committed to $150 million a month on compute.