Rumor has it that OpenAI is sitting on two important models:
1) Strawberry (formerly Q*), which does multi-step reasoning
2) GPT-5
If both are announced before march 2025, will GPT-5 come first? If both are not announced, the market will resolve N/A. Changes to their names, as long as the models in question are clearly the ones described, will not affect this market. If both get announced in the same sentence, or are actually the same model, resolves 50/50
Rumor that inspired this market:
People are murmuring that GPT-5 sometime in Sept-Oct, and this multi-step reasoning/L2 stuff in November
Update 2025-02-12 (PST) (AI summary of creator comment): Important Clarification:
The market cannot resolve to GPT-5 if Strawberry is announced before GPT-5.
If GPT-5 is not released before March 2025, the market will resolve as N/A.
Update 2025-02-12 (PST) (AI summary of creator comment): Clarification on GPT-5 Timing:
The event that counts for GPT-5 is its announcement, not its release.
If GPT-5 is not announced before March 2025, the market will resolve as N/A.
The market cannot resolve to GPT-5 if Strawberry is announced before GPT-5.
1,000It looks like this will probably be N/A.
Sam Altman on X: "OPENAI ROADMAP UPDATE FOR GPT-4.5 and GPT-5: We want to do a better job of sharing our intended roadmap, and a much better job simplifying our product offerings. We want AI to “just work” for you; we realize how complicated our model and product offerings have gotten. We hate" / X
@traders this market cannot resolve to GPT-5 since Strawberry was announced before GPT-5, BUT will resolve N/A if GPT-5 is not released before march 2025
@Bayesian Yeah, I'm not getting you. Clearly Strawberry was announced first since GPT-5 has not been announced yet.
Kinda baffled by how these question criteria were chosen but whatever, pulling out
@JoshYou Yes, it's certainly an 'interesting' criteria. I pulled out as well. I guess a reminder to read the criteria for every market, even when based on the title it seems impossible for vagaries.
My bad. I originally asked the question to see whether two different views (strawberry first then gpt-5, or gpt-5 first then strawberry) was more likely. as such I wanted to eliminate from the hypothetical the possibility that they didn't both get released. I realize now that wasn't worth the inconvenience it led to
@Bayesian Strawberry is released! OpenAI releases "Strawberry" ChatGPT model with better reasoning (axios.com)