Writing AI launched the consumer AI wave. ChatGPT's earliest adopters were bloggers and marketers. Jasper raised $125 million on the promise that AI would replace the copywriter. It was the obvious first win for the technology.

Our current scoreboard tells a different story.

Writing Sits at the Bottom

Jasper scores 7.6 in our latest rankings, the lowest score on the entire board. Its category peer, Notion AI, does better at 8.4, but even that trails Suno (music, 8.3), Midjourney (image generation, 8.2), and Higgsfield (video generation, 8.1).

Think about that for a moment. An AI music generator and a video tool outrank the category that was supposed to be the most natural fit for language models. Writing, the thing these models are literally trained to do, is finishing last.

Where the Top Actually Lives

Three of the five highest-scoring tools on our board are code editors: Cursor (9.0), Windsurf (8.5), and GitHub Copilot (8.4). This is not coincidence. Coding has a property that writing does not: verifiability. Code either compiles or it doesn't. Tests pass or they fail. The feedback loop is tight, improvement is measurable, and the value delivered per hour is obvious.

Cursor's 9.0 reflects a tool that genuinely changed how developers work, not one that saved them a few minutes on first drafts. Windsurf at 8.5 is competitive for the same reason. The discipline of software development gives AI something to anchor to.

Writing has no such anchor. A blog post is never "done" the way a function is. Quality is subjective. And every major writing tool now produces essentially the same output because they all draw from the same underlying models. Whatever differentiation Jasper once had has eroded to nothing.

The Commoditization Problem

Jasper's 7.6 is not a failure of execution. It is a structural problem. The moat was always shallow: a capable interface on top of a rented model. When OpenAI embedded generation into every surface, and when Notion built AI directly into its editor (at 8.4, it at least benefits from existing as a home for all your actual work), standalone writing tools lost their primary reason to exist.

Notion AI survives because the AI sits inside a product people already live in. Context is the product, not the prose.

What Holds Its Value

ElevenLabs at 8.6 is instructive here. Voice is a category where model output has a physical, perceptible quality: a synthesized voice sounds convincing or it does not. Perplexity AI at 8.6 survives because it built a search interface around AI rather than an AI product around text generation. Both are doing something harder to copy than "write me an article."

The shape of this scoreboard is essentially an argument: AI tools that replace a discrete, verifiable step in a real workflow hold their scores. AI tools that generate generic text into an undefined void do not.

Jasper's 7.6 is worth sitting with before the next writing AI startup pitches its Series A.