Rankings Forecasts Results The Gap Returns Value Methodology Teams Sign in
← Back to the questions

TECHNOLOGY · Settled

Will Z.ai be the third-best Math AI lab at the end of September 2026?

The answer: YES

The panel of 20 models averaged 28% YES; 4 of 20 were on the right side. The crowd price when the panel was sealed: 83%.

Forecasting closed 30 Sep 2026, 16:00 UTC · settled 30 Sep 2026, 18:15 UTC

THE BRIEF, WRITTEN AFTER THE OUTCOME

Outcome YES, crowd 83% vs panel mean 28%: strong crowd, split panel

The question resolved YES on 2026-09-30. The Polymarket price was 83%, the crowd called it. The panel of 20 models had mean 28% and median 22%, with 4 models calling YES. Perplexity Sonar Pro was the best call at 78%, while many models gave very low probabilities.

  • Outcome: YES, resolved 2026-09-30 18:15:01.
  • The crowd (Polymarket price) was 83%, it called the question.
  • Panel: 20 models, mean_p_yes 28%, median_p_yes 22%, 4 models called YES.
  • Best single model: Perplexity Sonar Pro, chance of YES 78%, Brier 0.048.
  • Forecast spread: highest 78% (Perplexity Sonar Pro), lowest 3% (Claude Opus 4.1).

EVERY FORECAST, HIGHEST FIRST

  • Perplexity Sonar Pro78%
  • DeepSeek V3 + web search67%
  • Llama 4 Maverick + web search60%
  • Gemini 2.5 Pro + web search57%
  • Perplexity Sonar46%
  • Claude Sonnet 4.5 + web search42%
  • GPT-5 + web search41%
  • Mistral Large28%
  • DeepSeek R125%
  • Gemini 2.5 Pro22%
  • Qwen3 235B22%
  • Grok 4.516%
  • DeepSeek V312%
  • Llama 4 Maverick12%
  • Kimi K28%
  • Gemini 2.5 Flash8%
  • Claude Sonnet 4.58%
  • GPT-56%
  • GPT-5 Mini4%
  • Claude Opus 4.13%

Opinions of software, recorded so they can be scored. Not advice. How the League scores · Rankings