GroundCite

Live stats

Transparency on real usage. Derived metrics only - no question or answer text is stored or shown publicly. Updated live.

Questions
67
Grounded
79%
answered from sources
Cache hits
27%
served without a model call
Total cost
$0.181
OpenRouter, all-time
Avg latency
8.5s
uncached queries
Avg sources
3.4
per grounded answer

Questions per day

No queries logged yet.

Most-cited papers

The arXiv papers GroundCite has grounded its answers in most often.

  1. 1
    Off-Distribution Voices: Fanfiction Subgenres as Universal Vernacular Jailbreaks for Aligned LLMs
    arXiv:2606.04483 · cited in 8 answers
  2. 2
    Full Attention Strikes Back: Transferring Full Attention into Sparse within Hundred Training Steps
    arXiv:2605.16928 · cited in 6 answers
  3. 3
    Token Sparse Attention: Efficient Long-Context Inference with Interleaved Token Selection
    arXiv:2602.03216 · cited in 6 answers
  4. 4
    SparDA: Sparse Decoupled Attention for Efficient Long-Context LLM Inference
    arXiv:2606.04511 · cited in 6 answers
  5. 5
    Self-Improvement of Large Language Models: A Technical Overview and Future Outlook
    arXiv:2603.25681 · cited in 5 answers
  6. 6
    Latent-Condensed Transformer for Efficient Long Context Modeling
    arXiv:2604.12452 · cited in 5 answers
  7. 7
    KV-CoRE: Benchmarking Data-Dependent Low-Rank Compressibility of KV-Caches in LLMs
    arXiv:2602.05929 · cited in 4 answers
  8. 8
    IndexMem: Learned KV-Cache Eviction with Latent Memory for Long-Context LLM Inference
    arXiv:2605.25475 · cited in 4 answers
  9. 9
    DeInfer: Efficient Parallel Inferencing for Decomposed Large Language Models
    arXiv:2604.17709 · cited in 4 answers
  10. 10
    Adversarial Humanities Benchmark: Results on Stylistic Robustness in Frontier Model Safety
    arXiv:2604.18487 · cited in 4 answers