Fulcrum's Echo Claims to Beat Frontier Models at Writing Style Imitation
davidad · x · 2026-10-02
Fulcrum says it evaluates its Echo model on style imitation—given the lack of good writing benchmarks—and reports it beats frontier models, even on authors unseen in training.
More from Models
- OpenAI Still Leads by a Solid Margin on Model Benchmark, Third-Party Notes — steipete · 2026-10-02
- Grok 4.7 now available on the Gemini Enterprise Agent Platform — SpaceXAI · 2026-10-02
- JevBench adds configurable cost tolerance: different budgets, different winners — airesearch12 · 2026-10-02
- No open models in top 25 anymore: MiMo #26, Qwen #28, GLM #30 — teortaxesTex · 2026-10-02
- Polymarket odds hit 88% for Claude Fable 5.2+ release by month-end amid grey-test leaks — Polymarket · 2026-10-02
- Hemmingway-1, a 27B human-like writing fine-tune, hits 32 tok/s on M5 Max and shines at roleplay — DerTomsn · 2026-10-02