llmledger

Every change to LiteLLM's model price catalogue since 6 September 2023: models added and removed, prices changed, retirement dates set. Also one row per model, and estimates of how likely a newly added model is to have its price changed, get a retirement date, or be removed.

Data as of 20 October 2026, 17:50 UTC (LiteLLM commit e63aec0), built with llmledger version 1.19: the same data its live page showed that day. The first upload (version 1.0 data, 13 October 2026) is in this repository's history.

• Code, method, specification and tests: https://github.com/pouriaetab/llmledger
• Live results, rebuilt every day: https://pouriaetab.github.io/llmledger/
• Archived release: https://doi.org/10.5281/zenodo.23330454

Not financial or procurement advice. The data describes edits to a public catalogue, not what any model costs you.

Files

changes.csv: one row per change (17,333 rows)

Column	Meaning
when	Time of the LiteLLM commit, UTC
sha	LiteLLM commit
model	Model name as LiteLLM lists it
kind	start (in the first version), added, removed, price (a field whose name contains cost or pricing changed), deprecation (the retirement date changed)
field	The field that changed, for price and deprecation rows
old, new	Value before and after, as JSON (null when absent)
flag	bulk (removed on a day with 50 or more removals), unit (a price made 100 times larger or smaller), reverted (undone within 30 days, or the change that undid it), likely (a likely correction: a price change in the model's first 7 days, from 0, or of 10 times or more either way), or empty

models.csv: one row per model (5,447 rows)

Model name, a grouped maker and name from naming rules (for example, eu.anthropic.claude-3-5-sonnet-20241022-v2:0 becomes anthropic / claude-3.5-sonnet), host, mode, status, dates added and removed, retirement date, input and output price in US dollars per million tokens, number of price changes (leaving out flagged ones), first exit, age in days for models still listed with no exit, and the cache read and cache write prices in US dollars per million tokens (the last two columns, added after version 1.0).

estimates.csv: competing risks estimates (100 rows)

Aalen-Johansen cumulative incidence of each first exit (price change, retirement, removal, none) at 30, 90, 180, 365 and 730 days after a model was added, with 95% bootstrap intervals (1,000 resamples) and the number of models still followed. Five definitions: main, headline (input and output token prices only), no_first_week, real_only (leaving out changes flagged bulk, unit or reverted), and likely_real (also leaving out likely corrections). In a labelled sample, about 6 in 10 first price changes were corrections of the catalogue rather than real price changes, so main overstates real ones.

How it was built

Every version of model_prices_and_context_window.json on LiteLLM's main branch was read in order with git and compared with the one before. The full method, definitions and known limitations are in the specification.

Citation

Emami Tabrizi, P. (2026). llmledger: what happens to an LLM after it joins a price catalogue (version 1.19.0). Zenodo. https://doi.org/10.5281/zenodo.23330454

Licence and source

The data is derived from BerriAI/litellm, which is available under the MIT licence below, and is shared under the same licence. llmledger is independent and not affiliated with or endorsed by BerriAI.

MIT License

Copyright (c) 2023 Berri AI

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.