New AI Tools to Try
in March 2026

Lovable lets you build full-stack web applications through natural conversation — powered by Claude. Describe what you want, and it builds a working app complete with a database, authentication, and deployment. No environment setup required. It's become a favourite for founders building MVPs quickly.

I tested this. Coming from a development background where spinning up a full-stack app involves significant configuration before you write a single line of product code, the speed of Lovable is striking. It is not for complex production systems — but for prototypes, MVPs, and internal tools, it delivers something working fast. The Claude foundation shows in the quality of the generated code structure.

Black Forest Labs' Flux models have made a serious dent in the AI image generation space. Flux.1 Pro produces photorealistic results that rival Midjourney, and the schnell (fast) variant is fully open source. Available via API and several third-party interfaces. Worth experimenting with if you generate images regularly.

OpenAI has been steadily improving GPT-4o's voice mode and real-time vision capabilities. The model can now maintain longer, more natural voice conversations and handle complex visual reasoning tasks. Free users get increased access to GPT-4o compared to earlier this year.

Anthropic has rolled out extended thinking mode for Claude, allowing it to reason through complex problems step-by-step before giving an answer. This significantly improves performance on maths, logic puzzles, and multi-step reasoning tasks. Available on Claude Pro.

As a daily Claude user, this is a meaningful update. The difference is most obvious on tasks that require holding multiple constraints at once — debugging a complex Unity script that interacts with several other systems, or planning a content structure that needs to balance multiple requirements. The visible reasoning process also helps you spot when the AI is going in the wrong direction early, before it delivers a long wrong answer.

DeepSeek released V3, continuing to close the gap with GPT-4 on coding and reasoning tasks at a fraction of the cost. The open-source weights allow running locally. Benchmark results are impressive — particularly for maths and code generation. Still free to use at chat.deepseek.com.

OpenAI's Sora text-to-video model is now available to ChatGPT Plus and Pro subscribers in more regions. Generates up to 60-second HD video clips from text prompts. Quality is impressive, though generation times are still slow and pricing is high for heavy use.

Gemini 2.0 Flash is Google's latest fast model — multimodal, real-time, and integrated deeply with Google Search and Workspace. It's shaping up to be the most capable free AI tool for users already in the Google ecosystem. The API is now available to developers.