# bert-base-german-cased vs bert-base-uncased: Specs and GPU Cost Source: https://savrn.com/models/compare/bert-base-german-cased-vs-bert-base-uncased Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- | Field | [bert-base-german-cased](https://savrn.com/models/bert-base-german-cased) google-bert/bert-base-german-cased | [bert-base-uncased](https://savrn.com/models/bert-base-uncased) google-bert/bert-base-uncased | | --- | --- | --- | | Publisher | BERT community | BERT community | | Task | Fill mask | Fill mask | | Modality | Text | Text | | Parameters, as reported | 110M parameters | 110M parameters | | Architecture | BertForMaskedLM | BertForMaskedLM | | Library | transformers | transformers | | Context length | 512 tokens | 512 tokens | | Repository size | 2.4 GB | 3.5 GB | | Artifact formats | safetensors, onnx, pytorch, jax, tf | safetensors, onnx, pytorch, coreml, jax, rust, tf | | License | mit | apache-2.0 | | Access | Open weights, no gate | Open weights, no gate | | Memory at 16-bit (weights and margin) | 0.3 GB | 0.3 GB | | Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 | | Memory at 4-bit (weights and margin) | 0.1 GB | 0.1 GB | | Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 | | Revision viewed | 0b061b4f7ce1 | 86b5e0934494 | | Downloads reported by the hub | 429.5k | 29.9M | | Last observed | 2026-09-18 | 2026-10-25 | An evaluation row appears only where at least two of these models report the same benchmark with the same stated configuration, metric, unit and setup. Different evaluators stay named in each cell. Values are shown as reported: no unit conversion, no ranking. ## SAVRN's Notes on bert-base-german-cased Fill-mask work in German is the job here: a 110M parameter encoder with a 30,000-token vocabulary and a 512-token window, evaluated on named entity recognition and classification. At 16-bit the weights are 0.2 GB and the model needs 0.3 GB, so the cheapest listed setup, one MI300X with 192 GB at $1.85 per hour, leaves nearly the whole card idle. In our facilities this rides beside other work; the question is how many copies you pack, not whether it fits. The MIT license permits commercial use, modification and redistribution as long as the copyright and permission notices travel with the files. Before committing, check that 512 tokens covers your documents, note that the publisher changed the vocabulary file in April 2020 and kept an old-vocab variant, and match your serving stack to one of the five shipped formats: safetensors, onnx, pytorch, jax or tf. SAVRN Research, 2026-09-18 ## SAVRN's Notes on bert-base-uncased Load this at 16-bit and it takes 0.3 GB of the 192 GB on the cheapest card in our table, one MI300X at $1.85 an hour on demand. Nobody buys a card for it; it rides on whatever is already running. The task it ships for is fill mask, the masked language modeling objective it was pretrained on with BookCorpus and Wikipedia, and 512 tokens is the most it takes in one pass. Apache 2.0 covers commercial use, modification and redistribution, provided you keep the license and copyright notices and state what you changed. Worth knowing before you commit: the team that released it never wrote a model card, so the card is secondhand, and the 16 files on disk total 3.45 GB across seven formats, safetensors to Rust, while the 16-bit weights you run are 0.2 GB. Pull one format, not the folder. SAVRN Research, 2026-09-18 ## Questions ### Which is larger, bert-base-german-cased or bert-base-uncased? bert-base-uncased (110M parameters) is larger than bert-base-german-cased (110M parameters), by the parameter counts their publishers report. ### Which is cheaper to run, bert-base-german-cased or bert-base-uncased? At 4-bit, bert-base-german-cased fits on 1x MI300X from $1.85 an hour and bert-base-uncased on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use bert-base-german-cased commercially? Yes. bert-base-german-cased is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included. ### Can I use bert-base-uncased commercially? Yes. bert-base-uncased is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ## Related Comparisons - [bert-base-cased vs bert-base-german-cased](https://savrn.com/models/compare/bert-base-cased-vs-bert-base-german-cased) - [bert-base-uncased vs camembert-base](https://savrn.com/models/compare/bert-base-uncased-vs-camembert-base) - [bert-base-uncased vs juribert-base](https://savrn.com/models/compare/bert-base-uncased-vs-juribert-base)