SAVRN Model Hub · Comparisons
gpt2 vs SmolLM2-135M
Gpt2 has 137M parameters and SmolLM2-135M has 135M parameters; gpt2 is released under MIT License and SmolLM2-135M under Apache License 2.0; at 16-bit, gpt2 needs about 0.3 GB (1x MI300X from $1.85 an hour) and SmolLM2-135M about 0.3 GB (1x MI300X from $1.85 an hour).
| Field | gpt2 openai-community/gpt2 | SmolLM2-135M HuggingFaceTB/SmolLM2-135M |
|---|---|---|
| Publisher | OpenAI community | Hugging Face Smol Models Research |
| Task | Text generation | Text generation |
| Modality | Text | Text |
| Parameters, as reported | 137M parameters | 135M parameters |
| Architecture | GPT2LMHeadModel | LlamaForCausalLM |
| Library | transformers | transformers |
| Context length | Not stated | 8,192 tokens |
| Repository size | 5.6 GB | 272.4 MB |
| Artifact formats | safetensors, onnx, pytorch, tflite, jax, rust, tf | safetensors |
| 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 | 607a30d783df | 93efa2f097d5 |
| Downloads reported by the hub | 15.4M | 2.2M |
| Last observed | 2026-09-18 | 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.
Other Reported Results
These results are listed for each model on its own, because the conditions needed to compare them are not stated or do not match. Two results that leave a condition blank are not assumed to share it.
SmolLM2-135M
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| TIGER-Lab/MMLU-Pro | Task mmlu_proMetric mmlu_proComparison conditions not established | 10.85 | EvalEval Reported by a third party |
Evaluated revision not stated | 2026-06-30 |
SAVRN's Notes on gpt2
Seven formats ship in this repository: safetensors, ONNX, PyTorch, TFLite, JAX, Rust and TensorFlow. That is why 26 files add up to 5,632,417,295 bytes for a 137M-parameter text generator, and why a mirror should hold only the format your serving stack loads. It was pretrained with a causal language modeling objective on raw English text, no human labeling. At 16-bit, weights and working memory are each 0.3 GB; on the cheapest setup we list, an MI300X with 192 GB at $1.85 per hour, the model is a rounding error and you size the card for whatever runs beside it.
MIT asks one thing in return for commercial use, modification and redistribution: the copyright and permission notices travel with every copy. Before committing, confirm the context window, which the listing does not state, and know that it records no reported evaluations and no per-token host price on the SAVRN Index.
SAVRN's Notes on SmolLM2-135M
Two trillion training tokens went into a model that fits in 0.3 GB. Hugging Face Smol Models Research trained this 135M-parameter SmolLM2, smallest of three sizes beside 360M and 1.7B, on FineWeb-Edu, DCLM and The Stack to run on-device. The cheapest priced card we list is one MI300X at $1.85 an hour, and 0.3 GB needed at 16-bit is why we would never dedicate one; here it shares a card, or goes to the device it was built for.
The only evaluation in our file is a third-party MMLU-Pro result of 10.85 from EvalEval, so run your own task first, and plan around an 8,192-token context; the training recipe is in arXiv:2502.02737. Apache 2.0 permits commercial use, modification and redistribution, requires keeping the license, copyright notices and any NOTICE file and stating significant changes, and carries an express patent grant from contributors, which matters if you fine-tune and ship.
Questions
Which is larger, gpt2 or SmolLM2-135M?
gpt2 (137M parameters) is larger than SmolLM2-135M (135M parameters), by the parameter counts their publishers report.
Which is cheaper to run, gpt2 or SmolLM2-135M?
At 4-bit, gpt2 fits on 1x MI300X from $1.85 an hour and SmolLM2-135M on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use gpt2 commercially?
Yes. gpt2 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 SmolLM2-135M commercially?
Yes. SmolLM2-135M 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.