Skip to main content
MANIFOLD
September 2026 AI model releases
61
Ṁ1kṀ14k
Sep 30
90%
Claude Opus/Sonnet/Haiku 5.x
64%
Kimi K3.x (or any Moonshot AI model)
58%
Minimax M3.x (or any Minimax model)
19%
TML-Large (or any Thinking Machines model)
15%
Gemini 3.x (or 4.x) Pro
9%
Deepseek V4.x Pro/Lite
Resolved
YES
Claude Fable 5.x
Resolved
YES
GPT-5.7 or GPT-6
Resolved
YES
Muse Spark 1.x (or any Meta model)
Resolved
YES
Gemini 3.x (or 4.x) Flash/Flash-Lite

This market resolves YES for each AI model that receives an official public release during August 2026. A release counts as available to users if it is widely made available through official channels (API, web interface, download, or official partnership). A US only release is enough to count as widely available.

Subvariants are model releases that share the same tier name (e.g., Flash Preview, Flash Live, Flash Instant, Flash Image all count for "Gemini 3.1 Flash"; 5.5-Thinking, 5.5-Instant, 5.5-Nano, 5.5-Pro all count for "GPT5.5"). Releases with a distinct tier name (Flash-Lite, Flash, Pro as a separate tier, etc.) do NOT count for an adjacent tier, even within the same model family. When in doubt, I'll use the official model name the company uses to classify it.

I will count any model release by OpenAI better than GPT-5.6 Sol for the GPT-5.7 option, even if they change the naming.

Previous Months:

/prismatic/august-2026-ai-model-releases

/prismatic/july-2026-ai-model-releases

/prismatic/june-2026-ai-model-releases

/prismatic/march-2026-ai-model-releases

/prismatic/april-2026-ai-model-releases

Get
Ṁ1,000
to start trading!
Sort by:
bought Ṁ180 YES

This resolves YES if GPT-6 is released, per the description. I renamed it for clarity.

@Fynn No, you need to N/A this. You can't do that

@mods plz

creator renamed the question and then bought all the yes votes

@0xseraphim The market description reads (since creation) "I will count any model release by OpenAI better than GPT-5.6 Sol for the GPT-5.7 option, even if they change the naming." I get that it is frustrating and I could have named the option better, but it would be worse if i changed it.

@Fynn Lol no you are going to N/A this. Cheating degrades the platform

@Fynn stop pretending that you're being fair and balanced. You forfeited that when you renamed the question and bought all the yes shares

@0xseraphim Do you agree that the new name of the option fits with the description of the market? If yes, then I don't see how renaming it would be a bad thing or "cheating". I agree that the inital name was not good, but it was an honest mistake.

@mods you need to intervene here

buddy either is intentionally dishonest or doesn't understand what he did was wrong. Which means he's almost certainly going to do this again

@0xseraphim I have reviewed the market description and even the trade ledger to see what happened here.

The crux of the argument is indeed the line from the description, which has always been there:

I will count any model release by OpenAI better than GPT-5.6 Sol for the GPT-5.7 option, even if they change the naming.

Under normal circumstances, if this line wasn't there, I would say that the renaming wouldn't be fair, but the "CYA" carve-out is right there. I can see how that could easily be missed if you're acting quickly and trading on the news where every second counts.

@Fynn Please consider constructing future markets in a way that is upfront and puts everything front-and-center "on the tin" for people who are moving quickly.

As it stands, though, I think we let this one ride.

@Quroe that ain't right. That means a user can just spam out markets like this then rename the questions halfway through to farm mana

@0xseraphim I sent you the 121 mana you invested as I don't want to profit off of my bad market naming. I think next time you could be more friendly/less angry, 121 mana is not that much and people on manifold generally operate in good faith.

@Fynn ok that's a lot more honest. But you could just have N/A'd the question as soon as the users pointed out the problem. That's what the N/A button is for

opened a Ṁ1,000 YES at 99.0% order

@0xseraphim no, we should never use the N/A button if the description was clear. N/A is BAD. It messes up a LOT with everyone's HAPPINESS. PLEASE do not ADVOCATE for N/A buttons if the DESCRIPTION was CLEAR. PLEASE PLEASE PLEASE.

121 mentioned