Resolves as YES if there is strong evidence that both OpenAI and Google DeepMind publicly release at least one new frontier image‑generation model each before December 1st 2025.

More precisely:

• Frontier image‑generation model

  • A new model (not available to the general public or typical enterprise/API customers before this market was created).

  • Clearly presented by the company as a flagship or “best yet” system for generating still images from text or multimodal prompts (for example, described in official materials as their latest or most advanced image model, or widely treated that way in coverage).

  • Represents a substantial qualitative improvement in image generation (e.g., fidelity, prompt adherence, text rendering, controllability) over that org’s previous public flagship, not just a trivial variant, style pack, or minor fine‑tune.

  • Can be either:

    • a dedicated text‑to‑image model, or

    • a multimodal model (like a future GPT/Gemini‑style system) where high‑quality image generation is one of the primary advertised capabilities.

• What counts as a “release”

  • The model must be made available to a broad set of external users, via at least one of:

    • Consumer products (e.g., ChatGPT, Gemini, web‑based image tools).

    • Developer access (e.g., OpenAI API, Gemini API, Vertex AI, or equivalent).

    • An open‑weight or downloadable release hosted on an official channel.

  • Limited or “paid preview” access does count as long as thousands of external users or developers can sign up or be onboarded in a reasonably standard way (not a tiny hand‑picked partner list).

  • A mere announcement, research paper, or stage demo without any real user or developer access does not count.

• What does not count

  • Purely video‑generation models with no distinct still‑image mode (e.g., a Veo‑style upgrade that only outputs videos) do not count.

  • Rebrands, small tweaks, pricing tiers, or purely performance/throughput optimizations of an existing image model do not count unless the company and credible coverage treat it as a genuinely new, more capable image model.

  • Internal-only deployments or heavily restricted private tests with a small number of hand‑picked partners do not count.

• “Both” requirement & resolution

  • The market resolves YES only if both:

    • OpenAI releases at least one qualifying frontier image‑generation model, and

    • Google/DeepMind releases at least one qualifying frontier image‑generation model,
      in each case strictly before 2025‑12‑01 (UTC).

  • If only one of the two organizations meets the criteria, or if neither does, the market resolves NO.

  • If there is borderline or ambiguous evidence for one side, the creator should err on the side of NO unless the new model clearly matches the criteria above and is widely regarded as a new flagship image generator.

The spirit of this market is to bet on whether both labs will push out another clearly frontier, broadly accessible image model in the short window leading up to December 1st 2025, not just incremental tuning of what already exists.