Apple is reportedly working on a "ChatGPT rival" project codenamed "Ajax" internally. Release here means an internal LLM made accessible to external consumers and developers in any shape or format. Could be an Xcode code helper, improvements to Siri, integrations to photo or notes app as long as an internally developed LLM is being used.
MOD ACTION
As this resolution was not clear cut, a panel of 3 uninvested moderators were asked to deliberate and decide this market's fate.
Our determination, by unanimous vote: NO.
Thanks for your patience, y'all. I'm now resolving this.
MOD ACTION
As this resolution was not clear cut, a panel of 3 uninvested moderators were asked to deliberate and decide this market's fate.
Our determination, by unanimous vote: NO.
Thanks for your patience, y'all. I'm now resolving this.
Apple released some models a couple of days ago, called OpenELM:
https://www.medianama.com/2024/04/223-apple-family-open-source-large-language-models-openelm/
Here’s What You Need To Know About Apple’s Family Of Open Source Large Language Models OpenELM
Apple has released OpenELM (Open-source Efficient Language Models), a family of eight open-source large language models (LLMs), on Hugging Face. The OpenELM family of models covers a parameter range of 270 million and 3 billion parameters (the weights and biases that a model learns during training). According to the company’s research paper on OpenELM, it outperforms Allen Institute for AI’s OLMo Models by 2.36% while requiring two times fewer pre-training tokens.
OpenELM consists of four pre-trained models and four instruct models. Pre-trained models are artificial intelligence (AI) models trained on large data sets to perform a specific task and can be further tuned for specific tasks. On the other hand, instruct models are AI models that have been fine-tuned to follow prompted instructions.
They are available for developers to download and use, here are the four instruct models:
https://huggingface.co/collections/apple/openelm-instruct-models-6619ad295d7ae9f868b759ca
These sound like small models appropriate for on-device use, and they're being framed as "proof of concept" rather than a product.
So I don't know whether they will count for this market, hence betting to 50%.
@Soli If it’s like a voice LLM-based chatbot then yes. Basically as long as a product which integrates an LLM developed internally (could be chatbot, summarizing, in Xcode, etc.) is made available to consumers and external developers.