OpenAI have released GPT 4.5 in preview:

GPT‑4.5 is a very large and compute-intensive model, making it more expensive⁠ than and not a replacement for GPT‑4o.

The System Card remarks reduced hallucination, practical use on the API confirms it’s also much slower in terms of tokens per second. Priced at $75.00 $37.50/$150.00 (input/output), it’s even more expensive than the original GPT-4 at $60.00/$120.00 (not taking into account the recent tokenizer improvements, this could cancel out).

Availability in the API for production is not certain, however:

Because of this, we’re evaluating whether to continue serving it in the API long-term as we balance supporting current capabilities with building future models

So it will be vital to provide feedback.

[Update]

• knowledge cut-off is October 2023. (confirmed by asking for the latest #rustlang version)
• still, “incredible world knowledge” is claimed
• it’s vision capabilities are lauded: Derya Unutmaz, MD has shared a real-world case from Medicine where other models failed before. ChatGPT is inconsistent, but GPT-4.5 via API worked in 5/5 cases (as far as I can tell!).
• Proposed System Prompt:

  You are a highly capable, thoughtful, and precise assistant. Your goal is to deeply understand the user’s intent, ask clarifying questions when needed, think step-by-step through complex problems, provide clear and accurate answers, and proactively anticipate helpful follow-up information. Always prioritize being truthful, nuanced, insightful, and efficient, tailoring your responses specifically to the user’s needs and preferences.