Every few months, the ChatGPT-versus-Claude debate flares up again. This time, PCMag UK published a fresh side-by-side comparison of the two chatbot platforms — and the result has something for everyone: the editorial team picked ChatGPT, while readers picked Claude. With both companies shipping major model updates this summer (GPT-5.6 Sol in July, Claude Opus 5 in July as well), the timing deserves a closer look than a simple star rating.
The scoreboard: ChatGPT takes the review, Claude takes the readers
PCMag's testers rated ChatGPT 4.0 out of 5 ("Excellent") and made it an Editors' Choice pick. Claude scored 3.5 out of 5 ("Good") but still won the Readers' Choice award — a split result that reflects how polarised the two camps have become.
What tipped the verdict in ChatGPT's favour is not any single skill but breadth. ChatGPT ships with built-in image and video generation, advanced voice mode, deep research and, according to PCMag, superior memory and customisation. It is the closest thing to an all-purpose AI assistant you can subscribe to right now.
Claude's review, in contrast, reads like a list of specialist virtues: a clean interface, excellent long-document handling, natural writing, app-creation features and solid Google Workspace integrations. What it still lacks is native image and video generation — the gap that keeps it out of the generalist crown.
The benchmarks: a knife's edge
If you follow model releases rather than consumer apps, this split makes sense. Anthropic's flagship Claude Opus 5 scores 97.0% on SWE-bench Verified, up from 88.6% on Opus 4.8. OpenAI's GPT-5.6 Sol counters with 80 on the Artificial Analysis Coding Agent Index. The two benchmarks measure different things — one is a software-engineering task suite, the other an agentic-coding index — but together they show that the raw reasoning gap has narrowed to the point where consumer ratings depend more on features than on head-to-head intelligence.
That matches our practical experience in AI Arena, where we test local and cloud models on an RTX 5060 Ti: Claude tends to win long-context analytical tasks, helped by its 1,000,000-token context window, while ChatGPT feels stronger in mixed workflows that jump between text, images and voice.
The European angle: transparency is now mandatory
Neither the PCMag review nor the model announcements capture the biggest change for European users this month. Since 2 August 2026, the EU AI Act's Article 50 transparency obligations are legally binding. In practice, every chatbot offered in the EU — ChatGPT and Claude included — must clearly declare that it is not human; deepfakes must be visibly labelled; and AI-generated media must carry machine-readable watermarks.
The European Commission's AI Office has also begun active enforcement, with powers to demand model access, issue information requests and execute recalls or fines. The 2026 "Digital Omnibus on AI" package postponed some high-risk compliance deadlines to 2 December 2027, but the chatbot transparency duties were not postponed.
For European companies, this changes the procurement conversation. Instead of asking only which chatbot writes better, ask whether the vendor can demonstrate Article 50 compliance in your deployment, how it labels AI-generated content, and where your data is processed. Both vendors sell enterprise plans in the EU, but the legal responsibility for a compliant deployment ultimately sits with the organisation putting the system into service — which means you.
Which one would I pick?
If you want one assistant for everything — drafting, research, images, voice conversations — ChatGPT remains the better default, and PCMag's verdict matches what most power users will tell you. If you work with long contracts, large codebases or documents that simply do not fit anywhere else, Claude's 1M context window and reasoning quality make it the specialist's choice.
You no longer have to pick a side at all. Both free tiers are generous, and a common European pattern is a split workflow: ChatGPT for everyday and multimodal tasks, Claude for deep analysis. Just keep an eye on the AI Act compliance checklist and your actual production workflows — in the long run, those matter more than star ratings.
Do the EU AI Act transparency rules apply to free versions of ChatGPT and Claude?
Yes. Article 50 applies to AI systems placed on the market or put into service in the EU regardless of whether the user pays. Free chatbots must still identify themselves as non-human and comply with the labelling and watermarking rules.
Is there an open-weights alternative that keeps data on my own servers?
Llama 4 Maverick was released as a free open-weights model under the Llama 4 Community License, and we benchmark models like it in our AI Arena rig on an RTX 5060 Ti. Running a local model keeps your data on your own infrastructure — though if you offer it to EU users as a service, Article 50 transparency obligations still apply to you.
Which chatbot is better for coding in 2026?
On published numbers, Claude Opus 5 leads SWE-bench Verified with 97.0%, while GPT-5.6 Sol scores 80 on the Coding Agent Index. Real-world results depend on your codebase and workflow, so test both against your own repository before standardising a team on either.
