For the first time since ChatGPT’s launch, Claude takes the lead in the French AI user satisfaction ranking. Across all categories, ChatGPT keeps first place on usage volume, but loses 7 NPS points in one quarter. Breakdown in six numbers.
How the Study Was Conducted
1,200 French respondents, recruited via Comparateur-IA and an external panel. All use at least one generative AI tool at least once per week. Sample weighted for age, gender, occupation, and region.
Measures: declared usage (multiple responses), frequency, NPS, reasons for satisfaction and dissatisfaction (qualitative coding), payment intent.
Usage in France in March 2026
Ranking by declared usage share (at least once / week):
| Tool | Usage Share | Change Q+1 |
|---|---|---|
| ChatGPT | 78 % | −2 pts |
| Claude | 54 % | +9 pts |
| Gemini | 41 % | +6 pts |
| Copilot Microsoft | 33 % | +4 pts |
| Mistral / Le Chat | 22 % | +3 pts |
| Perplexity | 19 % | +2 pts |
The NPS Shift
This quarter’s news is Claude’s position at the top of NPS – the first time since we’ve been measuring.
“With Claude, I don’t feel like I’m fighting a consumer product anymore. The tone is better, refusals are rarer and more justified.”
- +58
Claude
Leading on French writing, long reasoning, code.
- +51
ChatGPT
Versatility, ecosystem, but perception of ‘more standardized’.
- +29
Gemini
Strong progress with Workspace, held back by consumer UX.
- +24
Copilot
Appreciated in Office, but confined to office productivity.
- +21
Mistral / Le Chat
Praised for sovereignty, awaited on quality.
Who Wins on Which Front?
📝Writing & Editing
Winner: Claude
Style, nuance, tone, fidelity to long instructions.
💻Code
Winner: Claude / GPT-5 (preview)
Claude solid on review and refactor, GPT-5 awaited on generation.
🖼️Images
Winner: ChatGPT / Midjourney
ChatGPT for integrated use, Midjourney for raw quality.
🎙️Voice
Winner: ChatGPT (voice mode)
Strong advantage thanks to advanced voice mode.
🔎Search
Winner: Perplexity / Gemini
Sourced answers, real-time web access.
📊Office Productivity
Winner: Copilot / Gemini
Native integration into Office suite / Workspace.
Newsletter
Receive AI Analysis, Once a Month
Insights like this one, no hype, no spam.
Three Standout User Personas
The Pragmatic Solo
Freelancer, independent. ChatGPT as main tool, adds Perplexity for search. AI budget: €20-40/month.
The Team Operations
Employee at SME/mid-size company. Integrated Copilot + Claude for writing. Progressive switch to Mistral for compliance.
The Technical Power-User
Dev, data, product. Heterogeneous stack: Claude (writing, code), ChatGPT (image, voice), Cursor (IDE). Budget €60-100/month.
What We Anticipate Through September
- 1
Saturation of the Consumer Market
Adoption pace is flattening. Growth shifts to depth of usage and spending.
- 2
Shift to Paid
Intention to pay exceeds 60% among respondents using AI daily.
- 3
Multiplication of Stacks
Share of users with ≥ 3 monthly tools rises from 41% to expected 50%.
- 4
Sovereignty in Question
Demand for FR/EU alternatives rises, but real usage share stays constrained.
- 5
Arrival of GPT-5
Potential wildcard: a successful GPT-5 could reshuffle the deck in one quarter.
Frequently Asked Questions
How Is Satisfaction Calculated? +
Standard NPS (1-10) on the question: ‘Would you recommend this tool to a colleague?’ NPS = % promoters (9-10) − % detractors (0-6). 1,200 French respondents active on generative AI in 2025-2026.
Why Does Claude Overtake ChatGPT in France? +
Three reasons recur: French writing quality (style and nuance), long-document handling, refusals perceived as less arbitrary. ChatGPT keeps advantage on images (GPT image) and voice.
Is Gemini Growing? +
Yes, especially among Google Workspace users: Docs/Gmail/Drive integration and ultra-long context. Gemini’s NPS climbs +18 in six months but stays below 30, held back by perception of a consumer product.
Where Does Mistral Stand in France? +
Mistral is progressing in enterprise and public sector, supported by sovereign anchoring. In consumer, the usage share stays modest (8%) despite good privacy perception.
How Many Respondents Use Multiple Tools? +
73% of respondents use at least two tools per month. The ‘AI stack’ practice is generalizing: a main assistant, a code IDE, an image generator, a voice/video tool.
Also Read
⚠️ Transparency: some links may be affiliate. No impact on our ratings or pricing.