SAVRN Model Hub · Comparisons
bert-base-cased vs bert-base-german-cased
Bert-base-cased has 109M parameters and bert-base-german-cased has 110M parameters; bert-base-cased is released under Apache License 2.0 and bert-base-german-cased under MIT License; at 16-bit, bert-base-cased needs about 0.3 GB (1x MI300X from $1.85 an hour) and bert-base-german-cased about 0.3 GB (1x MI300X from $1.85 an hour).
| Field | bert-base-cased google-bert/bert-base-cased | bert-base-german-cased google-bert/bert-base-german-cased |
|---|---|---|
| Publisher | BERT community | BERT community |
| Task | Fill mask | Fill mask |
| Modality | Text | Text |
| Parameters, as reported | 109M parameters | 110M parameters |
| Architecture | BertForMaskedLM | BertForMaskedLM |
| Library | transformers | transformers |
| Context length | 512 tokens | 512 tokens |
| Repository size | 1.8 GB | 2.4 GB |
| Artifact formats | safetensors, pytorch, jax, tf | safetensors, onnx, pytorch, jax, tf |
| License | apache-2.0 | mit |
| 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 | cd5ef92a9fb2 | 0b061b4f7ce1 |
| Downloads reported by the hub | 3.6M | 429.5k |
| Last observed | 2026-10-09 | 2026-09-18 |
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-cased
Nobody needs to size a server around 109 million parameters, and the memory line says why: at 16-bit the weights come to 0.2 GB and the model runs in 0.3 GB. The cheapest Index setup, one MI300X with 192 GB at $1.85 an hour on-demand, is the smallest slice we list, not what it asks for. Run it on what you own; it fills masked tokens in English text inside a 512 token window, and capitalization changes its answer.
Apache 2.0 lets you ship it in a commercial product if the license and notice files travel with it and you state significant changes. Check that 512 tokens covers your inputs, that bookcorpus and wikipedia pretraining fits your domain, and that a 2024-02-19 last update is recent enough. The repository runs to 1.8 GB across ten files, holding safetensors, pytorch, jax and tf, so pull the format your stack loads.
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.
Questions
Which is larger, bert-base-cased or bert-base-german-cased?
bert-base-german-cased (110M parameters) is larger than bert-base-cased (109M parameters), by the parameter counts their publishers report.
Which is cheaper to run, bert-base-cased or bert-base-german-cased?
At 4-bit, bert-base-cased fits on 1x MI300X from $1.85 an hour and bert-base-german-cased on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use bert-base-cased commercially?
Yes. bert-base-cased 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.
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.