Decision Index rebalances weights: category share scales with sqrt of benchmark count
multimodalart · x · 2026-09-26
multimodalart details the new weighting in Decision Index 0.2.1: previously every category had equal share, so areas with fewer benchmarks carried disproportionate weight (1/n each). Now each core area's share scales with the sqrt(n) of its benchmark count, giving each benchmark a 1/n weight — a middle ground between equal-category and equal-benchmark weighting. Arts & taste is fixed at 10% since it's less central to practical applications.
Related event: Decision Index 0.2.1 Fixes SGD Benchmarks, Revamps Category Weights(2 posts)→
More from Models
- Opus 5.5 does worse and costs more at Max thinking — Medium beats Max on hard benchmark — davidyin44 · 2026-09-26
- LibertAI ships open-weight Deem 9B on Qwen3.5, trailing Jev 68.9% vs 74.1% — Pokenhagen · 2026-09-26
- Peter Steiberger says he codes with Codex plus an OC harness, not Claude — steipete · 2026-09-26
- Why Anthropic lags OpenAI on math: compute constraints and 400k GPUs coming online — haider1 · 2026-09-26
- Hytale WorldGen V2 face-off: Claude Opus 4.6 vs Opus 5.5 compared — Angaisb_ · 2026-09-26
- OpenAI's rumored persistent agent "o" may tie to old "rebranding to O" report — Dullydude · 2026-09-26