Speculation: V4 Pro Could Hit 93.0 on Terminal Bench if It Matches Flash's Gains
ChrisGPT · x · 2026-08-06
Twitter user ChrisGPT extrapolates V4 Pro benchmark scores based on V4 Flash's improvements from Preview to 0731, predicting Terminal Bench 93.0, CyberGym 90.7, and DeepSWE 59.9, emphasizing it's pure speculation.
More from Models
- Meta's Spark Model Priced Below Competitors, Aiming to Acquire Training Data — teortaxesTex · 2026-08-06
- Google Explains Gemma E2B/E4B: How PLE Boosts Power Without Adding Parameters — GlennCameronjr · 2026-08-06
- Muse Spark Team Announces Major Improvements in Coding Capabilities — alexandr_wang · 2026-08-06
- Prime Agent Coding Harness Tops ARC-AGI-3 with 95.5% Beating Human Experts — xeophon · 2026-08-06
- Report: Ilya's SSI Model to be a Small Reasoning Engine, Outperforming Fable — iruletheworldmo · 2026-08-06
- Meta Offers Near-Free AI Models For Training Data, Disrupting Google's Strategy — scaling01 · 2026-08-06