What will be true of Llama 4? [Add answers]
15
1.6kṀ2813
Aug 16
80%
Some version of it will score better on GPQA than o1-preview (73% pass@1)
71%
It will be able to output audio
64%
Some version of it will score better on SWE-Bench Verified than Claude 3.5 Sonnet (October version)
24%
It will be able to output images
16%
Gets the high score on Humanity's Last Exam
14%
Manifold will think any variant is the best model released up to that point in a poll
10%
Gets the high score on FrontierMath
7%
Reaches Highest Arena Score on Chatbot Arena
Resolved
YES
It will have a variant with over 400 billion parameters (does not need to release immediately)
Resolved
YES
It will have a variant with over 1 trillion parameters (does not need to release immediately)

if an answer doesn't specify which model, any model in the llama 4 family meeting the criteria causes it to resolve YES

high score options are up to date of release - if llama4 gets a high score on april 1 and gpt-9000 comes out on april 2 and beats its score, answer still resolves YES

  • Update 2025-04-05 (PST) (AI summary of creator comment): Important Update:

    • Llama 4 Behemoth has not yet been released.

    • Answers that rely solely on Llama 4 Behemoth will remain unresolved until its release, unless smaller models in the Llama 4 family satisfy the criteria.

  • Update 2025-05-30 (PST) (AI summary of creator comment): The market is being reopened because Llama 4 Behemoth is not expected to be released soon. This action is consistent with the previous update that answers relying solely on Llama 4 Behemoth will remain unresolved until its release.

Technology
AI
LLMs
Grab Bag
What Will Happen Markets
Get
Ṁ1,000
to start trading!


Sort by:
19d

reopening as behemoth does not seem like it is going to be released anytime soon

1mo

@mods trying to resolve NO

Error resolving answer: RangeError: Variable $2 out of range. Parameters array length: 1 at /usr/src/app/node_modules/pg-promise/lib/formatting.js:188:19 at String.replace (<anonymous>) at Object.array (/usr/src/app/node_modules/pg-promise/lib/formatting.js:172:22) at Object.formatQuery (/usr/src/app/node_modules/pg-promise/lib/formatting.js:296:29) at Task.$query (/usr/src/app/node_modules/pg-promise/lib/query.js:129:40) at Task.<anonymous> (/usr/src/app/node_modules/pg-promise/lib/query.js:275:23) at Task.query (/usr/src/app/node_modules/pg-promise/lib/task.js:117:34) at Task.obj.multi (/usr/src/app/node_modules/pg-promise/lib/database.js:922:30) at getContractAndMetricsAndLiquidities (/usr/src/app/backend/shared/src/utils.ts:173:28) at Task.<anonymous> (/usr/src/app/backend/shared/src/resolve-market-helpers.ts:87:50)

1mo

@SaviorofPlant weird, same for me. @ian ? (or more reliable to post on discord bugs)

29d

@SaviorofPlant fixed, sorry!

2mo

Is this market meant to be closed?

2mo

It's out: https://ai.meta.com/blog/llama-4-multimodal-intelligence/

Llama 4 Behemoth has not yet released, most answers will not resolve until it does (unless some of the smaller models satisfy the criteria).

reposted 3mo

this looks like it might be the next big LLM release (or maybe deepseek r2), for those of you looking for dopamine after this insane month

© Manifold Markets, Inc.•Terms•Privacy