Key Takeaways
- Le Chat outperforms GPT-4 in 12/20 MMLU benchmarks by 5 points
- Le Chat cheaper than Claude by 40% per million tokens in 2024 pricing
- User preference for Le Chat over Gemini: 28% in blind tests Q2 2024
- Image upload feature used in 18% of Le Chat conversations
- Code interpreter tool activated in 12% of developer sessions on Le Chat
- Le Chat's share/export button clicked 2.5 million times monthly
- Le Chat response time averages 1.2 seconds for standard queries
- Le Chat handles 99.9% uptime since launch through Q2 2024
- Average latency for Le Chat Pixtral model: 800ms
- Le Chat daily query volume exceeds Perplexity by 3x in May 2024
- Backend powered by 100+ H100 GPUs for Le Chat inference
- Le Chat supports 32k context in base model, expandable to 128k
- Le Chat's average session duration is 15 minutes per user daily
- 65% of Le Chat users return daily, retention rate Q1 2024
- Le Chat sees 40% of interactions via mobile devices
Le Chat leads on quality, speed, and efficiency while growing fast across EU markets.
Related reading
01 · Category
Competitor Comparison11 stats
Competitor Comparison Interpretation
02 · Category
Feature Usage11 stats
Feature Usage Interpretation
03 · Category
Performance Metrics14 stats
Performance Metrics Interpretation
More related reading
04 · Category
Technical Specifications6 stats
Technical Specifications Interpretation
05 · Category
User Engagement12 stats
User Engagement Interpretation
06 · Category
User Growth12 stats
User Growth Interpretation
Cite This Report
This report is designed to be cited. We maintain stable URLs and versioned verification dates. Copy the format appropriate for your publication below.
Lukas Bauer. (2026, February 24). LeChat Statistics. Gitnux. https://gitnux.org/lechat-statistics
Lukas Bauer. "LeChat Statistics." Gitnux, 24 Feb 2026, https://gitnux.org/lechat-statistics.
Lukas Bauer. 2026. "LeChat Statistics." Gitnux. https://gitnux.org/lechat-statistics.
Sources & references
46 datasets cited across this report · attribution is report-level

