What are the hardware requirements for Deepseek Coder 6.7B Base?

Deepseek Coder 6.7B Base requires approximately 13.5 GB of VRAM and supports a context window of 16K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Deepseek Coder 6.7B Base and how large is it?

Deepseek Coder 6.7B Base is developed by deepseek-ai, a model with 6.7b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Deepseek Coder 6.7B Base perform on standard benchmarks?

Deepseek Coder 6.7B Base has the following published scores: MMLU 38.39, HumanEval (Python) 45.83. Compare against reference models on this page or on the LLM Explorer leaderboards.

Deepseek Coder 6.7B Base by deepseek-ai — VRAM 13.5GB, 16K context

Name: Deepseek Coder 6.7B Base
Rating: 3.5 (2 reviews)
Author: deepseek-ai

Deepseek Coder 6.7B Base is an open-source language model by deepseek-ai. Features: 6.7b LLM, VRAM: 13.5GB, Context: 16K, License: other, Code Generating, LLM Explorer Score: 0.19, Arc: 37, HellaSwag: 53.5, MMLU: 38.4, GSM8K: 18, HumanEval: 45.8.

Codegen Endpoints compatible Llama Pytorch Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/deepseek-ai/deepseek-coder-6.7b-base

Deepseek Coder 6.7B Base Parameters and Internals

Model Type

code completion, code insertion, repository level code completion

Use Cases

Areas:

code development, software engineering

Primary Use Cases:

project-level code completion, code infilling

Additional Notes

Offers various sizes from 1.3B to 33B parameters.

Supported Languages

English (code and natural language), Chinese (code and natural language)

Training Details

Data Sources:

code, natural language

Data Volume:

2 trillion tokens

Context Length:

16000

Model Architecture:

Multi-Head Attention

LLM Name	Deepseek Coder 6.7B Base
Repository 🤗	https://huggingface.co/deepseek-ai/deepseek-coder-6.7b-base
Model Size	6.7b
Required VRAM	13.5 GB
Updated	2026-07-15
Maintainer	deepseek-ai
Model Type	llama
Model Files	10.0 GB: 1-of-2 3.5 GB: 2-of-2 10.0 GB: 1-of-2 3.5 GB: 2-of-2
Generates Code	Yes
Model Architecture	LlamaForCausalLM
License	other
Context Length	16384
Model Max Length	16384
Transformers Version	4.34.1
Tokenizer Class	LlamaTokenizerFast
Beginning of Sentence Token	<｜begin▁of▁sentence｜>
End of Sentence Token	<｜end▁of▁sentence｜>
Vocabulary Size	32256
Torch Data Type	bfloat16

Rank the Deepseek Coder 6.7B Base capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Quantized Models of the Deepseek Coder 6.7B Base

Model	Likes	Downloads	VRAM
Deepseek Coder 6.7B Base GGUF	23	19460	2 GB
Deepseek Coder 6.7B Base AWQ	10	16513	3 GB
Deepseek Coder 6.7B Base GPTQ	5	20	3 GB
Deepseek Coder 6.7B Base AWQ	0	0	3 GB

Best Alternatives to Deepseek Coder 6.7B Base

Best Alternatives	Context / RAM	Downloads	Likes
...Deepseek Coder 6.7B Energy Sft	16K / 13.5 GB	261	0
Speechless Coder Ds 6.7B	16K / 13.5 GB	459	7
...s Coder6.7b Reflct Adamw Iter1	16K / 13.5 GB	475	0
...Coder6.7b Reflct Rmsprop Iter1	16K / 13.5 GB	95	0
...Coder6.7b Reflct Rmsprop Iter1	16K / 13.5 GB	110	0
...r6.7b Pos Reflct Rmsprop Iter1	16K / 13.5 GB	87	0
...ir4 Ds Coder6.7b Rmsprop Iter1	16K / 13.5 GB	43	0
...r6.7b Pos Reflct Rmsprop Iter1	16K / 13.5 GB	90	0
Code 290K 6.7B Instruct	16K / 13.5 GB	87	6
...Coder6.7b Reflct Rmsprop Iter1	16K / 13.5 GB	62	0

Note: green Score (e.g. "73.2") means that the model is better than deepseek-ai/deepseek-coder-6.7b-base.