I will explain the strengths of each AI out there that I use, so you can know exactly which AI models/tools are the best in my opinion! This is what these AIs are good at, in my opinion! Text Generation (LLMs): -Claude 4.5 Sonnet by Anthropic (Perplexity): •Questions in general •Web search •Vision -GPT-5 by OpenAI (Perplexity): •Writing -GPT-5 Thinking by OpenAI (Perplexity): •Reasoning (STEM) -GPT-5 Codex (Cursor): •Coding Image Generation: -Seedream 4.0 by ByteDance (Perplexity): •Best overall -Midjourney V7 by Midjourney: •Best aesthetic Video Generation: -Veo 3.1 Fast by Google (Perplexity): •Best overall Voice Generation: -GPT-4o (AVM) by OpenAI (ChatGPT): •Best overall Music Generation: -Suno V 4.5 by Suno AI: •Best overall If I haven’t mentioned an AI, it’s either because I’ve forgotten about it or it’s useless for me. Changelog: -I added GPT-5, Claude 4.5 Sonnet and Seedream 4.0 to the list.

Midjourney just released Niji V7, their latest AI anime image generation model. This time, Niji supports video generation, specifically fine-tuned for creating anime scenes.

Hailuo AI just released Minimax M2.1, an agentic open-source model that is great at coding. You can use it in Cline.

Gemini 3 is now available in the Gemini CLI (AI coding agent) for all users (even free users).

My Cursor Usage:

Google released Gemini 3 Flash, a fast, cheap, and good AI model (LLM). It's nearly as good as premium models like Gemini 3 Pro, but Google offers unlimited usage for free on Gemini with Gemini 3 Flash.

Google will release Gemini 3 Flash soon. It's going to be cheap, fast and good at many tasks.

OpenAI is currently rolling out its new image generation model for ChatGPT users.

Runway has just released the best AI video generation model, Gen-4.5, to everyone on a paid plan.

Mistral AI released two new coding models. Devstral 2 (123B) under a modified MIT licence, and Devstral Small (24B) under Apache 2.0. They also released a native CLI.

Google will release Gemini 3 Flash soon. It's currently getting tested in LM Arena.

Ben just released Halo. Halo is a habit app that has an AI coach that helps you reach your goals. It's currently only available on iOS.

FAL AI now has Seedream 4.5, an AI image generator. It's really cheap at 0.04 USD.

Runway has just released the best AI video generation model, Gen-4.5. It's a lot better at creating high-quality videos and following instructions.

DeepSeek AI just released DeepSeek-V3.2 and DeepSeek-V3.2 Speciale. The Speciale version competes with Gemini 3 Pro (Best AI model right now) for a much cheaper price.

Gemini can now make PDFs. Just select Canvas and tell it to generate a PDF.

OpenAI is planning to start testing ads on ChatGPT soon (we can deduce it from the ad-related strings). It's probably going to only affect free users.

Kimi AI can now create presentations by searching the web and creating slides. It's currently free for a limited time.

Perplexity now lets you see how clothes would look on you (AI image editing) before buying them. The try-on feature is available to all paying Perplexity users.