Analytics
Computed live from approved public records.
Sampling bias: The Chat Archive is self-selected. People choose which conversations to share, and imported datasets reflect their own collection methods (for example, WildChat users of a free ChatGPT proxy). These numbers describe this archive, not AI use in general. Model names for community contributions are mostly self-reported.
Conversation growth (last 26 weeks, by publish week)
No data yet.
Model distribution
No data yet.
Provider distribution
No data yet.
Topic distribution
No data yet.
Language distribution
No data yet.
Source distribution
No data yet.
Conversation length & feedback
- Avg. user turns
- 0.00
- Avg. messages
- 0.00
- Avg. characters
- 0
- Single-turn
- —
- 2–4 turns
- —
- 5+ turns
- —
- User corrected model
- —
- Verified evaluation
- —
“User corrected model” is a heuristic text match (for example “that’s wrong”). It is a browsing aid, not a validated label.
Contribution activity (30 days)
No community submissions in the last 30 days.
Dataset growth (releases)
No releases yet.