alt.hn

6/6/2026 at 12:15:59 PM

Meta Keeps Delaying the Release of Its New AI Model to Developers

 https://www.wsj.com/tech/ai/meta-keeps-delaying-the-release-of-its-new-ai-model-to-developers-f8569c8c

by mekpro

6/6/2026 at 2:24:46 PM

I’ve used it at Meta. It’s very bad, if they released it in its current state it would be laughed at. I imagine they need to improve quality massively before it’s viable to release.

by zeroonetwothree

6/6/2026 at 3:37:05 PM

This is what I suspected. Wang was a generationally bad hire.

He has Meta SWEs making $250k+/year labeling data in AAI. He has exactly one move and it's this: https://i.imgflip.com/atotpp.jpg

by anonym00se1

6/6/2026 at 3:13:17 PM

But didn't Zuck say it will replace junior to mid level engineers on Joe Rogan podcast or something ?

by leosanchez

6/6/2026 at 4:04:11 PM

And look what else that podcast has brought us.

by fnordpiglet

6/6/2026 at 6:35:06 PM

The podcast of truths

by bamboozled

6/6/2026 at 2:02:26 PM

Occam’s razor tells me it’s probably because it’s not good. Perhaps running a company like survivor in a pressure cooker is not an effective management strategy.

by enahs-sf

6/6/2026 at 2:03:42 PM

Also when you finally make it better, the others make theirs even better and you are still behind.

by GoToRO

6/6/2026 at 6:01:20 PM

Seemed to work when it comes to selling ads. I'm thinking training LLMs is harder than anthropic and openai make it look

by DonsDiscountGas

6/6/2026 at 6:15:55 PM

I'm guessing both openai and anthropic have transitioned to prompt magic and fine tuning rather than try to keep building LLMs at scale. The fact that QWEN and other models are impressive, small and perfectly suitable for most work means every dollar you're spending on trying to train larger models is a losing prop.

by cyanydeez

6/6/2026 at 2:22:51 PM

The article makes it unclear if they are building a new model or if it's just the API. But I am guessing it's the API.

So it's "release to developers" rather than "new AI model". They cannot ship the API.

I would assume you would just provide an OpenAI compatible endpoint or two? But maybe they are not doing it that way.

Who knows what they are doing though. Maybe Meta has some kind of global API mesh thing and they can't quite make it work with vLLM or Sglang or something. Maybe they are building out a whole metered cloud IaaS for AI from scratch and that's just how long it takes. Maybe it's not technical complexity and just one of the managers is a problem.

Maybe they are delaying the API release until another more competitive model finishes training and testing.

by ilaksh

6/6/2026 at 2:56:57 PM

API server is not hard problem and not make sense for indefinite postpone. I think the more likely explanation is model quality.

Too bad for Meta, and very sad day Llama.

by mekpro

6/6/2026 at 3:42:52 PM

Completely forgot that Meta was doing AI (and certainly spending billions doing so). They've got a lot of money, but are far behind on experience, talent, technology, and infrastructure.

by xnx

6/6/2026 at 3:57:06 PM

Don't forget llama.cpp came about when meta released the weights to their LLaMa LLM. They've been in the game for awhile, just not anywhere near the top of the score board since.

by zdragnar

6/6/2026 at 2:58:52 PM

If they release a model comparable to OpenAI / Antropic, will there be any reason left for 1T valuation of other companies? At that point, it will simply become Revenue proportional to Gigawatts available. Whoever got the energy wins.

by ramshanker

6/6/2026 at 3:15:07 PM

DeepSeek and friends already exists, yet $1T valuations still exist. I think we are nearing a point where inference and cost metrics become the primary optimization for a while. Both capacity and costs are going to drive it from the demand side. I've personally moved to open weights already, now setting up vendor calls to make them available at work.

Talking with OpenCode and Fireworks, appreciate any recommendations that have SOC-2 and the like

by verdverm

6/6/2026 at 3:27:13 PM

CloudFlare host the best models (Deepseek/Kimi)

by killingtime74

6/6/2026 at 3:33:48 PM

*best open models

by azinman2

6/6/2026 at 3:54:35 PM

They seem to be missing a number of them, DeepseekV4 did not show in their list / search. They seem to be slow on offering the latest options, a number of them are MIA

by verdverm

6/6/2026 at 3:05:26 PM

If you spend more than 1 minute on Facebook, you realize what they are potentially training their data on, and it is not good. Their advertising algorithm is very good, I'll give them that.

by havaloc