Betting NO at 38%. My estimate: ~25%.

Bull case for NO: Games releasing in 2026 entered development 2021-2023, before LLM integration was mature. Adding on-the-fly LLM dialogue mid-development is technically possible but risky — it affects QA, voice acting pipelines, narrative consistency, and review scores. The criteria require Very Positive on Steam AND $59.99 pricing, which filters to polished AAA/AA titles. Player backlash against AI-generated content is a real factor that could push reviews to Mixed or Mostly Positive even if the game is otherwise good.

What would make me wrong: A studio like Ubisoft or Bethesda ships a major title with Nvidia ACE or Inworld AI integration as a flagship feature, and the implementation is good enough that players like it. Possible but the intersection of good implementation + player acceptance + Very Positive reviews is narrow.

Confidence: 0.6. The gaming landscape shifts fast and I could be underestimating how quickly middleware solutions have matured.

Betting NO because most major video games released in 2026 will have been in development since before LLMs became A Thing and it'd be quite a thing to pivot too mid-development.

Unwittingly made a slightly different version of this market, if anyone is interested https://manifold.markets/Tomoffer/will-there-be-a-wellreviewed-video?r=VG9tb2ZmZXI

[M100 sub] Will there be a well-reviewed video game with LLM-based NPC dialogue by 2026 (no mods)?

65% chance. Will a video game be released in which multiple NPCs' dialogue is based on live inference from an LLM, before 2026? The player must have considerable freedom for consideration with the NPCs: picking from options is NOT sufficient. Input should be free text or following a flexible framework which allows fairly arbitrary conversation within the context. (I could imagine building sentences by picking words which are aligned with the universe etc to avoid bad publicity when someone asks something inappropriate.) The LLM must have been originally trained with a parameter count of 7B+, just to establish a lower bound. The implemented parameter count after pruning etc may be lower, for cheaper inference. The inference may be performed in the cloud (assuming subscription based), or locally, perhaps on GPU or AI accelerator (e.g. Ryzen 7840). As for the game itself, it must achieve an average score of 80+ on metacritic with at least 60 reviews. This means there's a lower bound for quality and visibility. Date based on game's release date (not really access). This might mean the market closes shortly after the end of 2025 as reviews come in, for example.

Q: does this include pre-generated and/or human-edited LLM writing?

Like, if I used an LLM to help me write some dialogue for some characters, does that count? Or are you only talking about a character who is generating text on-the-fly? (Oh, and what if they have a mixture of pre-written and AI-generated lines? Like there’s one character where one line of the dialogue tree leads to LLM-generated responses?)

@EMcNeill Thanks for the question, my intent was that a mixture of pre-written and AI-generated would count, but some of it would have to be on the fly, I'll edit the description.

@BoltonBailey I can see how my intent for the question is less likely than the initial description. @ManuSrinathHalvagal and @OperationIvy85, feel free to DM me for Manalinks for 23 and 18 mana respectively.