Continuously monitoring 316 active endpoints

LLM API outputs are unstable over time.

LLM APIs are opaque black boxes, even for open-weight models. TrackLLM continuously monitors them for undisclosed changes — cheaply — with logprob tracking (LT) where logprobs are exposed, and black-box border input tracking (B3IT) where they aren't. Small changes are fine, but they should be disclosed.

Endpoints monitored
984
316 active · 668 retired · 1,915 in the catalog
Models tracked
279
across 55 orgs
Providers
83
207 serving variants
Queries logged
15.7M
since Jun 2025
Changes detected
250
79 LT · 171 B3IT

Inria IRISA CNRS SequoIAQ
Subscribe

Latest changes

how far each endpoint moved from its baseline All 250 changes →
13 Oct
Probe-prompt outputs shifted (total variation 0.21)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.42)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.34)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.54)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.34)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.46)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.30)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.45)
B3IT
10 Oct
Probe-prompt outputs shifted (total variation 0.31)
B3IT
9 Oct
Probe-prompt outputs shifted (total variation 0.24)
B3IT
More…

Providers

detected changes per endpoint-year of LT monitoring All 83 providers →
More…

Endpoints

984 endpoints monitored, of 1,915 in the catalog · 250 changes across 194 endpoints All endpoints →
Endpoint Status Changes Methods Last change Drift from baseline
deepseek-v4-flash-0731streamlake/fp8changed 3 LTB3IT 9 Sep
trinity-large-thinkingarcee-aichanged 3 B3IT 22 Sep
claude-sonnet-4.5claude-on-awschanged 2 B3IT 9 Sep
deepseek-v3.2atlas-cloud/fp8changed 2 LTB3IT 10 Oct
deepseek-v4-flashparasail/fp8changed 2 B3IT 3 Oct
glm-5.2gmicloud/fp8changed 2 LTB3IT 29 Aug
glm-5.3-flashcloudflarechanged 2 B3IT 4 Oct
glm-5.3-flashdigitaloceanchanged 2 B3IT 3 Oct
gpt-4o-2024-05-13azurechanged 2 LT 21 May
gpt-4o-miniazurechanged 2 LTB3IT 23 May
More…

Timothée Chauvin, Erwan Le Merrer, François Taïani, Gilles Tredan (2026), "Log Probability Tracking of LLM APIs". ICLR 2026. https://arxiv.org/abs/2512.03816

Timothée Chauvin, Clément Lalanne, Erwan Le Merrer, Jean-Michel Loubes, François Taïani, Gilles Tredan (2026), "Token-Efficient Change Detection in LLM APIs". ICML 2026. https://arxiv.org/abs/2602.11083

@inproceedings{
chauvin2026log,
title={Log Probability Tracking of {LLM} {API}s},
author={Timothee Chauvin and Erwan Le Merrer and Francois Taiani and Gilles Tredan},
booktitle={The Fourteenth International Conference on Learning Representations},
year={2026},
url={https://openreview.net/forum?id=hFxivbAgVP}
}

@inproceedings{
chauvin2026tokenefficient,
title={Token-Efficient Change Detection in {LLM} {API}s},
author={Timothee Chauvin and Cl{\'e}ment Lalanne and Erwan Le Merrer and Jean-Michel Loubes and Francois Taiani and Gilles Tredan},
booktitle={Forty-third International Conference on Machine Learning},
year={2026},
url={https://openreview.net/forum?id=7cMlZZYZT0}
}