316 results on one curve: ECI scaling plots align GSM8K through FrontierMath scores
burny_tech · x · 2026-11-21
burnytech shared a 'From ECI to benchmark scores' chart aggregating 316 results, placing model scores on GSM8K, OTIS Mock AIME 2024-2025, and FrontierMath Tiers 1-3 (v2) onto a single ECI (Epoch Capabilities Index) axis, producing clean S-shaped scaling curves.
- Labeled models include Qwen2.5-Coder-7B, Qwen-14B, o3-mini, GPT-4.1 nano, o1, Grok 4.6, and Inkling-Small
- Shows how well a single capability index predicts scores across benchmarks
- Accompanied by an 'If you would please consult the graphs' meme jabbing at takes that ignore the data
More from Models
- Unverified claim: Riemann hypothesis solved, empty google-deepmind/math-horizon repo spotted — imjustnewatai · 2026-11-21
- Anthropic mathematician says he and Claude proved the Riemann Hypothesis — a 250,000-page proof — rohanpaul_ai · 2026-11-21
- Dev argues transformers can't separate instructions from context, enabling jailbreaks — gerardsans · 2026-11-21
- AI week in numbers: Anthropic's $93.5B IPO, 250k-page RH proof draft — TheTuringPost · 2026-11-21
- HeyGen unveils Decision 1: ~100ms multimodal reasoning model, claims 4x faster than Gemini — HeyGen · 2026-11-21
- NVIDIA Cosmos 3 hands-on: world model forgets the room after 8 seconds off-camera — PrateekVJoshi · 2026-11-21