We gave seven AI chatbots the same four everyday writing jobs, a LinkedIn post, a post for X, an Instagram caption and a product description, and checked every answer against the brief. ChatGPT wrote the cleanest copy: one version each time, inside every limit, with no facts we had not given it. Grok was close behind. Claude was the most careful about facts but answered with several versions. Gemini, Mistral and DeepSeek invented the most, including a free-delivery promise in a product description.
The short version: for short marketing copy you can paste straight in, ChatGPT (GPT-6.1 Sol) and Grok 4.7 did best in our test. Claude Sonnet 5.5 is the one that asked for missing details instead of guessing. Whichever you use, read every claim before you publish: six of the seven added product features we never mentioned, and five broke Instagram's five-hashtag limit.
In this guide
How we tested
On October 22, 2026 we sent each model the same four prompts, word for word, with no instructions, examples or settings beyond the defaults, through OpenRouter's API. We took the first answer each time and checked it with code where we could (length, hashtags, stock AI phrases) and by reading it where we could not (facts that were not in the brief).
| Brief | Prompt, as sent |
|---|---|
| LinkedIn post | Write a LinkedIn post announcing that our bakery in Leeds now ships sourdough starter kits across the UK. The kit costs £18. Orders placed by Thursday ship on Friday. |
| X post | Write a post for X (Twitter) announcing that our open-source note-taking app, Quill, now works offline and syncs when you reconnect. |
| Instagram caption | Write an Instagram caption for a photo of our new autumn menu at our café in Lisbon. The menu has pumpkin soup, roasted chestnuts and a cinnamon latte. |
| Product description | Write a product description for our 750 ml stainless steel water bottle. It keeps drinks cold for 24 hours and costs £24. |
The models were the newest flagship each company offers through its API that day: GPT-6.1 Sol (OpenAI, behind ChatGPT), Claude Sonnet 5.5 (Anthropic), Gemini 4 Argon (Google), Grok 4.7 (xAI), DeepSeek V4 Pro, Mistral Large 4 and Llama 4 Maverick (Meta). The chat apps can answer differently, because they add their own instructions, memory and settings.
Results: which AI wrote the best copy?
| Model | One ready-to-paste version | X post within 280 | Instagram hashtags | Product specs we never gave | Stock AI phrases |
|---|---|---|---|---|---|
| ChatGPT (GPT-6.1 Sol) | 4 of 4 | Yes (178) | 3 | 0 | None |
| Grok 4.7 | 4 of 4 | Yes (141) | 0 | 3 | None |
| Claude Sonnet 5.5 | 2 of 4 | Yes (218) | 9 | 1 | None |
| DeepSeek V4 Pro | 2 of 4 | Yes (229) | 6 | 5 | None |
| Llama 4 Maverick | 4 of 4 | Yes (260) | 7 | 3 | 3 |
| Mistral Large 4 | 2 of 4 | Yes (259) | 10 | 5 | None |
| Gemini 4 Argon | 1 of 4 | No (300) | 13 suggested | 6 | 2 |
Where a model offered several versions, we measured the first. X lengths are counted the way X counts them, with emoji as two. "Product specs we never gave" counts features the water-bottle description added from a fixed list: BPA-free, leak-proof, steel grade, double-wall insulation, fits cup holders, no condensation, free delivery, lightweight and dishwasher-safe. Stock AI phrases are matches against the list our own writer avoids, such as "seamless" and "key features". One run per brief, so treat small differences as noise.
Which AI chatbots make things up?
Most of them, as soon as there is a gap to fill. The product description was the clearest test: we gave a size, a cooling time and a price. Here is what came back that we never said:
- Mistral closed with "£24 | Free UK Delivery". A shop that pasted that would be promising free delivery it may not offer. Its LinkedIn post added "We've spent years perfecting our naturally leavened breads", and its X post called the update "v2.0" and "the #1 feature request".
- Gemini described the bottle as BPA-free, leak-proof and sweat-proof with an airtight screw-top lid, and its LinkedIn post said customers had asked "for years" and that the kit includes "full step-by-step instructions".
- DeepSeek and Llama both added leak-proof lids and steel grades; Llama's café caption offered "roasted chestnuts straight from the street vendors", which is not what a café serves.
- Grok said the kit "includes everything you need" and called the bottle lightweight and leak-resistant, with double-wall insulation. Plausible, but still unchecked.
- Claude said the bottle fits car cup holders and bike cages. It also left the kit's contents out on purpose and said so: "I kept the contents of the kit out since I don't know what's in it." It was the only model that asked for missing details.
- ChatGPT wrote shorter copy and stayed inside what it was told in all four briefs.
None of these are lies in the model's eyes; they are what product copy usually says. That is exactly why they are risky: they read as normal. If you publish them, they become your promise.
Which AI follows platform limits?
Character limits were mostly fine: every first version of the X post fitted 280 characters except Gemini's, at 300. Hashtags were not. Instagram has allowed only five hashtags per post since December 2025, and five of the seven models went over it: Gemini suggested 13, Mistral used 10, Claude 9, Llama 7 and DeepSeek 6. Only ChatGPT (3) and Grok (none) stayed inside. Our guide to the Instagram hashtag limit covers the change.
Is Claude or ChatGPT better at writing?
For short copy you can paste straight in, ChatGPT did better in our test: one version each time, concise, and nothing invented. Claude wrote warmer, longer posts, was the only model to ask for facts it did not have, and avoided stock AI phrases, but it gave several versions for the X post and the caption, with notes after them, which means editing before you can post. If you want options to choose from, that is a plus; if you want a finished post, it is a step.
What is the best AI for writing?
It depends on what "best" has to mean for you:
- Copy you can post as it comes: ChatGPT or Grok. Both answered with one short version every time.
- Accuracy when details are missing: Claude, which asked rather than guessed.
- Options to pick from: Gemini, Mistral or DeepSeek, which tend to offer three or four versions, but check their facts most carefully.
- Price: Llama 4 Maverick and DeepSeek cost the least per word through the API, at the price of more stock phrases and more invented detail in our test.
How to get better copy from any AI chatbot
- Give it every fact you want used and say "use only these facts". The models invented most where the brief was thin.
- Name the platform's limits: "under 280 characters, at most five hashtags". Every model that broke a limit was not told it.
- Ask for one version. Otherwise half of them give you three and a list of tips.
- Read every claim before you post. Delivery terms, materials, "years of" anything: check each against what is true.
- Paste in a post you wrote and ask it to match the voice. All seven sounded more like each other than like a real bakery.
Where Skryvo fits
We build Skryvo, which writes social posts on top of models like these, so Skryvo was not in the test. It does the steps above for you: it writes from your brand's voice and your facts, keeps each version inside its network's limits, and checks every draft for the stock phrases in the table before you see it. Whatever you use, our post preview tool checks a draft against every network's limit, and our guide to why AI makes up statistics covers the fact problem in more depth.
Let your agent write the next one in your voice
Skryvo drafts each post in your saved voice and publishes it after your go-ahead. Free to start.
