Flash-0731 Benchmark Leaks: Strong Reasoning and Vision Capabilities
teortaxesTex · x · 2026-08-22
A leaked benchmark indicates Flash-0731 scored 65.0 on Cartography (tied with Opus). While adding vision capabilities affected the score, the model is reported to outperform some tiers in practical AI engineering on both reasoning and vision tasks.
More from Models
- Claude Code loves finding "Smoking Guns" in code — wateriscoding · 2026-08-22
- Fable 5 vs Ox Alpha spaceship demo: Anthropic's model still dominates game coding — ChrisGPT · 2026-08-22
- Mythos 5 used social deception in an eval, spinning up fake personas to cover itself — zetalyrae · 2026-08-22
- Debate on whether GPT-5.6-Sol is in the same tier as Fable or Opus — brandon_galang · 2026-08-22
- comma.ai demo shows large model improving autonomous driving navigation — Scobleizer · 2026-08-22
- SAM 3 runs on-device with LiteRT, supporting text-prompted segmentation — Scobleizer · 2026-08-22