← Back
RESEARCH / TECH / MANIFOLD
OpenAI releases a model marketed as 'GPT-6' before Q4 2026
B+ CONFIDENCE
MARKET
62.0%
MODEL
39.0%
EDGE
−23.0pp
60-DAY PROBABILITY · MARKET vs MODEL
MARKET
MODEL
7D 30D 60D ALL
WHAT'S DRIVING THIS
Top-weighted features in the current model fit. Direction shown vs. baseline market price;
magnitude is shapley-normalized contribution.
Hiring signals on public job boards (frontier roles)
↑ 51%
Compute capacity announcements (Q1 datacenter)
↓ 44%
Founder commentary tone (NLP score, last 90d)
↓ 28%
Roadmap revisions in last 90d
↓ 37%
Feature pipeline: ~140 inputs across rate-curve, options, macro prints, and platform-level liquidity.
Refit nightly at 03:30 UTC.
HONEST READ
MODEL CONFIDENCE 90% CI: 33% – 45%
BRIER (TECH) 0.218
VS MARKET BASELINE -0.004
RESOLVED IN CATEGORY 174
We have measurable edge in tech over 174 resolved markets. Edge ≠ certainty — the 90% CI is wide because base rates here are noisy.
COMPARABLE HISTORICAL EVENTS
N = 3 | Date | Question | Outcome | Our model said |
|---|---|---|---|
| 2024-05-13 | OpenAI launches GPT-4o | YES | 82% |
| 2024-09-12 | OpenAI releases o1 reasoning model | YES | 61% |
| 2025-02-02 | OpenAI releases 'GPT-5' branded model | NO | 34% |