New AI Models should be added: Grock and Phi-3
Amos Glenn
xAI's and Microsoft's latest LLMs should seriously be added to the list of premium models. Additionally, more models should be added for different kinds of content creation to work with audio, 3D models, and eventually video. Plenty of models like Point-E exist to create 3D models, and models could be used with audio for voice cloning and music creation. It's an idea, and the technology exists, widely available.
J
Joel Midden
Merged in a post:
Add Qwen3 and GPT-4.5-Preview Model Support
MG “MMGG” minGGe
Please consider integrating the new Qwen3 models (Qwen3-32B, Qwen3-30B-A3B, Qwen3-235B-A22B) and also adding support for GPT-4.5-Preview. These top-tier models would enrich user choice and boost platform competitiveness. Please consider.
J
Joel Midden
Merged in a post:
[Models List] Considering Adding Codestral to the Offering
J
Julien
Hi there,
I’d like to request the addition of Codestral 25.01 to your models list. It was released in April 2025, supports over 80 programming languages, has 124 billion parameters, and seems to excel in code generation and document analysis. You can find more information at https://mistral.ai/news/codestral-2501
Thanks!
J
Joel Midden
Merged in a post:
[Models List] Considering Adding Pixtral Large to the Offering
J
Julien
I wanted to drop a quick note about something I mentioned a while back. I had requested to add Pixtral Large to your offerings when this model launched, but it looks like my ticket got closed as "resolved" due to a mix-up with the Mixtral models. Pixtral Large has been told a "game-changer": with 124 billion parameters, it tends to excel at handling tasks like document analysis and image understanding. Plus, its 128K context window lets it work with multiple high-res images seamlessly (told to handle 30 Hi-res pics seemlessly) . Compared to other top models, it shows nice results on benchmarks like MathVista and DocVQA. I know the list is already long, but adding Pixtral Large could be great for the community! Thanks for considering this.
Kindle Humpback
This: Qwen3
I don't mind so much about the GPT-4.5 - would have preferred to have these as two separate tasks. But of course all advanced models are welcome.