Open-weight model
DeepSeek-R1-Distill-Llama-70B-oQ5e
by Matias Almeida maptan/DeepSeek-R1-Distill-Llama-70B-oQ5e
DeepSeek-R1-Distill-Llama-70B-oQ5e is an open-weight model from Matias Almeida. It has 70.6B parameters and a 131,072-token context. At 16-bit it needs about 169.3 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 DeepSeek-R1-Distill-Llama-70B-oQ5e (70.6B 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 | 141.1 GB | 169.3 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x MI325X $2.00 · 1x MI355X $2.59 |
| 8-bit | 70.6 GB | 84.7 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x MI325X $2.00 · 1x MI355X $2.59 |
| 4-bit | 35.3 GB | 42.3 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.
DeepSeek-R1-Distill-Llama-70B-oQ5e 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 Matias Almeida.
Configuration
- Architecture
- LlamaForCausalLM
- Context length (tokens)
- 131,072
- Layers
- 80
- Hidden size
- 8,192
- Feed-forward size
- 28,672
- Attention heads
- 64
- Key/value heads
- 8
- Head dimension
- 128
- Vocabulary size
- 128,256
- RoPE base
- 500000
- Stored precision
- bfloat16
- Model type
- llama
Identity and Version
- Repository
- maptan/DeepSeek-R1-Distill-Llama-70B-oQ5e
- Publisher
- Matias Almeida
- Task
- Not stated by the source
- Modality
- Other
- Library
- mlx
- Parameters
- 70.6B parameters
- Languages
- mlx, oq
- Revision
- 3dd55dd55d8429fcc6a4f4c97bcb9fbaf0d8707f
- First published
- 2026-10-02
- Last updated
- 2026-10-02
Files and Weights
18 files, 50.3 GB in total. The weights are 10 files totalling 50.3 GB in safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model-00001-of-00010.safetensors | Weights | 5.1 GB | 97bfdfe6c2d1 |
| model-00002-of-00010.safetensors | Weights | 5.1 GB | f9f2db49f432 |
| model-00003-of-00010.safetensors | Weights | 5.1 GB | 616b0dbac622 |
| model-00004-of-00010.safetensors | Weights | 5.0 GB | 0a2ce15fb366 |
| model-00005-of-00010.safetensors | Weights | 5.1 GB | 7ad5858b0423 |
| model-00006-of-00010.safetensors | Weights | 5.1 GB | 31acf8cfbf00 |
| model-00007-of-00010.safetensors | Weights | 5.0 GB | 235d4468ed64 |
| model-00008-of-00010.safetensors | Weights | 5.0 GB | aa93e13a0050 |
| model-00009-of-00010.safetensors | Weights | 5.1 GB | 6b5a4aa541e6 |
| model-00010-of-00010.safetensors | Weights | 4.4 GB | 4003e2dedce6 |
| config.json | Configuration | 13.0 KB | — |
| generation_config.json | Configuration | 181 B | — |
| model.safetensors.index.json | Configuration | 150.9 KB | — |
| oq_imatrix_report.json | Configuration | 24.5 KB | — |
| README.md | Documentation | 621 B | — |
| .gitattributes | Repository | 1.5 KB | — |
| tokenizer.json | Tokenizer | 9.1 MB | — |
| tokenizer_config.json | Tokenizer | 3.1 KB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 50.3 GB
Released by Matias Almeida through its official repository on Hugging Face.
Built From
- Derived from deepseek-ai/DeepSeek-R1-Distill-Llama-70B
- Quantized from deepseek-ai/DeepSeek-R1-Distill-Llama-70B
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 50.3 GB |
| 16-bit | 141.1 GB |
| 8-bit | 70.6 GB |
| 4-bit | 35.3 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About DeepSeek-R1-Distill-Llama-70B-oQ5e
How much GPU memory does DeepSeek-R1-Distill-Llama-70B-oQ5e need?
About 169.3 GB at 16-bit and 42.3 GB at 4-bit: the weights (70.6B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run DeepSeek-R1-Distill-Llama-70B-oQ5e 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 DeepSeek-R1-Distill-Llama-70B-oQ5e's context length?
131,072 tokens, from the maximum position embeddings in its published configuration.