Datasets
The corpora behind our models.
Collections10
Largest150K rows
LicenseOpen
Newest releases first: pretraining tokens, reasoning traces, prompt-routing sets, and chat-title collections. Names link to Hugging Face.
| Datasets (all 10) | Samples | Downloads | Likes |
|---|---|---|---|
| Supra2-100M-tokens | — | 58+ | 2 |
| LLM-self-identification | 459 | 358+ | 12 |
| agentic-reasoning-trace-summaries-40k | 43.7K | 312+ | 9 |
| Prompt-Routing-Dataset | 992 | 529+ | 19 |
| reasoning-summaries-61k | 61K | 627+ | 19 |
| supra-wild-titles-130k | 134K | 318+ | 4 |
| chat-titles-filtered-115K | 116K | 550+ | 8 |
| chat-titles-unfiltered-150K | 150K | 505+ | 3 |
| chat-titles-12K | 11.9K | 325+ | 3 |
| SupraThink-Dataset-500x | 500 | 459+ | 11 |