d1-3B-oQ8e is an open-weight model from Towa. It has 3.1B parameters and a 32,768-token context. At 16-bit it needs about 7.5 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
This model was quantized using oQ (oMLX v0.7.0) mixed-precision quantization.
Runs On
What it takes to serve d1-3B-oQ8e (3.1B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 6.2 GB | 7.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 3.1 GB | 3.7 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 1.6 GB | 1.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 18, 2026.
d1-3B-oQ8e on every accelerator the SAVRN Index prices, at every precision
Model Card
This model was quantized using oQ (oMLX v0.7.0) mixed-precision quantization.
Excerpt from the card by Towa.
Configuration
- Architecture
- Lfm2VlForConditionalGeneration
- Context length (tokens)
- 32,768
- Layers
- 30
- Hidden size
- 2,048
- Feed-forward size
- 10,752
- Attention heads
- 32
- Key/value heads
- 8
- Vocabulary size
- 128,000
- Model type
- lfm2_vl
Identity and Version
- Repository
- RepublicOfKorokke/d1-3B-oQ8e
- Publisher
- Towa
- Task
- Not stated by the source
- Modality
- Other
- Library
- mlx
- Parameters
- 3.1B parameters
- Languages
- mlx, oq
- Revision
- 2231f9e43f105b9eb7a363d78195b775cf0e6ae4
- First published
- 2026-10-08
- Last updated
- 2026-10-08
Files and Weights
15 files, 3.7 GB in total. The weights are 1 file totalling 3.7 GB in safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 3.7 GB | 2b1c6462cc02 |
| api.py | Configuration | 3.5 KB | — |
| config.json | Configuration | 2.9 KB | — |
| hybrid.py | Configuration | 14.4 KB | — |
| lfm2_vl.py | Configuration | 5.6 KB | — |
| modeling_d1.py | Configuration | 1.3 KB | — |
| oq_imatrix_report.json | Configuration | 11.9 KB | — |
| processor_config.json | Configuration | 824 B | — |
| prompt.py | Configuration | 12.5 KB | — |
| runner.py | Configuration | 11.8 KB | — |
| README.md | Documentation | 328 B | — |
| chat_template.jinja | Other | 5.4 KB | — |
| .gitattributes | Repository | 1.6 KB | — |
| tokenizer.json | Tokenizer | 17.9 MB | 8096ecb9f545 |
| tokenizer_config.json | Tokenizer | 507 B | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 3.7 GB
Released by Towa through its official repository on Hugging Face.
Built From
- Derived from LiquidAI/d1-3B
- Quantized from LiquidAI/d1-3B
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 3.7 GB |
| 16-bit | 6.2 GB |
| 8-bit | 3.1 GB |
| 4-bit | 1.6 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About d1-3B-oQ8e
How much GPU memory does d1-3B-oQ8e need?
About 7.5 GB at 16-bit and 1.9 GB at 4-bit: the weights (3.1B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run d1-3B-oQ8e on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What is d1-3B-oQ8e's context length?
32,768 tokens, from the maximum position embeddings in its published configuration.