Best LLMs for Topic Organization & Clustering — DTP Benchmark
Discover natural groupings without external schema, name and order them coherently.
Discover natural groupings without external schema, name and order them coherently.
Task-by-task breakdown
ps_section_reassignment autogenerated
No model has reached MEDIUM confidence yet — accumulating evidence.
Topic Discovery Clustering (pooled)
Pooled TT for thematic topic discovery from a batch of items: topic_clustering_batch (content-driven) and topic_clustering_claims (claim-driven). Same output schema, same capability.
| Rank | Model | Quality | Cost / call | vs best |
|---|---|---|---|---|
| 1 | Claude Opus 4.7 best | 8.77 | $0.305172 | — |
topic_cluster_naming autogenerated
| Rank | Model | Quality | Cost / call | vs best |
|---|---|---|---|---|
| 1 | Qwen 3.6 Plus | 8.20 | $0.004549 | 87% |
| 2 | Haiku 4.5 | 8.22 | $0.006792 | 81% |
| 3 | Kimi K2.6 | 8.50 | $0.012598 | 64% |
| 4 | Claude Sonnet 4.6 | 8.20 | $0.020378 | 42% |
| 5 | DeepSeek V4 Pro | 8.27 | $0.021492 | 39% |
topic_clustering_assign_sections autogenerated
| Rank | Model | Quality | Cost / call | vs best |
|---|---|---|---|---|
| 1 | Qwen 3.5 Flash | 8.65 | $0.000085 | 99% |
| 2 | DeepSeek V4 Flash | 8.40 | $0.000337 | 95% |
| 3 | Qwen 3.6 Plus | 8.73 | $0.000865 | 87% |
| 4 | Kimi K2.6 | 8.64 | $0.002421 | 64% |
| 5 | Gemini 3.1 Pro Preview | 8.61 | $0.002663 | 60% |
topic_sequence_determination autogenerated
No model has reached MEDIUM confidence yet — accumulating evidence.