In 2025, Gemini's overall performance improved significantly. This year, Gemini was consistently the most mentioned model in almost every discussion about basic models. Gemini 3.0 Thinking, in particular, is highly recommended. Its multimodal capabilities, extensive contextual understanding, world knowledge, and powerful search and understanding abilities make it the most powerful model in terms of overall performance. It also comes bundled with the top-tier drawing software Nano Banana Pro and Veo 3.1 – one membership for three top-tier features. It offers excellent value for money.
I primarily use Gemini for structured outputs in my daily work, such as writing frameworks and corporate training frameworks, as well as for research and multimodal collaboration.
It remains very effective at "dialogue-based progression" in handling complex problems. Especially when you want to gradually refine a vague idea into an actionable solution, its conversational experience is excellent. Another point that everyone values is memory. GPT has accumulated many of my long-term projects and preferences, enhancing the sense of continuous collaboration. This is particularly advantageous when writing long-term series of articles, delivering consulting work, and producing repetitive outputs. Furthermore, the writing experience with GPT-5.2 Thinking is indeed quite good.
For text-based conversations, I use Gemini as the primary platform, while ChatGPT is used for in-depth discussions and supplementary functions.
In my experience, if you take content creation, marketing, and commercial delivery seriously, you'll eventually need a tool that consistently produces high-quality visuals. My go-to tool is Nano Banana Pro. It's integrated with Gemini; you can use it once you buy a Gemini membership, without having to pay separately.
I use it to create many of the images, infographics, flowcharts, indicator diagrams, cover images, and more for my blog, all thanks to its powerful modeling capabilities.
Veo 3.1: Has advantages in visual language and certain textural aspects. Kling AI: Best suited for filmmaker-oriented AI video generation. Runway: Best suited for flexible creation and advanced editing. Luma AI's Dream Machine focuses on realistic video generation. OpenAI's Sora continues to showcase impressive technical capabilities. Because there are so many AI video generator models now, I found a platform that aggregates them all, namely Viddo AI. The purpose is to allow me to choose different models based on the requirements of the video I want to create. Importantly, with a membership to this platform, I can use virtually all models available on the market. In addition, this platform also has image generation models such as Nano Banana Pro, GPT 4o, and Flux.1 Kontex. Very comprehensive.
AI Video Generator & AI Image Generator
If you're more focused on details like the naturalness, emotion, and timbre of the voice, then TTS (Text-to-Speech) is definitely worth exploring. Voice is a very subtle thing. Once you've experienced natural TTS, you can't go back to "robot-like" voices.
Tools for creating narration, courses, voiceovers, and branded audio.
Musical structure is inherently highly modular. With the integration of AI, the "cost of composing music" can be reduced to a very low level. If I had to recommend one, it would be Suno. Suno's strength lies in its ability to create professional-quality songs in just one minute, regardless of genre or timbre. Without close inspection, you simply cannot tell that it was generated by AI.
AI Music Generator
Claude Code is one of the most perfect AI programming products of 2025. It has driven the practical application of AI programming, including innovative designs such as MCP and Skills, all proposed by the Claude team. Using Claude Code as an AI coding agent yields excellent results, showcasing many brilliant design details.
AI Programming