OLMo 1 models
2 OLMo 1 models from Ai2 on Hugging Face, from 1.2B to 6.9B parameters, published by Ai2 in F32, with a 2K-token context. The smallest official model, OLMo-1B-hf, needs about 5.3 GB of VRAM at its published precision; the cheapest live fit is V100 at $0.088/hr.
Part of the OLMo series · Next generation: OLMo 2
Pick a size
One row per official OLMo 1 size: the VRAM it needs at each precision, the cheapest GPU that holds it today, and the KV cache for a 32K-token context.
| Model | Parameters | Native VRAM | FP8 VRAM | INT4 VRAM | Live GPU fit (native) | Est. $/hr | KV cache at 32K |
|---|---|---|---|---|---|---|---|
| OLMo-1B-hf | 1.2B | 5.3 GB | 1.3 GB | 0.7 GB | V100 | $0.088/hr | 4.0 GB |
| OLMo-7B-hf | 6.9B | 30.8 GB | 7.7 GB | 3.8 GB | V100 | $0.187/hr | 16.0 GB |
VRAM is weights times a flat 1.2 overhead; the KV cache is a separate per-model figure at 16-bit, shown where the architecture is published. See the methodology.
Official models (2)
| Model | Parameters | Published as | VRAM needed | Live GPU fit | Est. $/hr |
|---|---|---|---|---|---|
| OLMo-1B-hf | 1.2B | F32 | 5.3 GB | V100 | $0.088/hr |
| OLMo-7B-hf | 6.9B | F32 | 30.8 GB | V100 | $0.187/hr |
VRAM is for the precision the model is published in: the weights times a flat 1.2 overhead, with the KV cache not included. See the methodology. The fit shown is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
