What are the hardware requirements for Moonlight L3 15B 16K?

Moonlight L3 15B 16K requires approximately 30 GB of VRAM and supports a context window of 8K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Moonlight L3 15B 16K and how large is it?

Moonlight L3 15B 16K is developed by Darkknight535, a model with 15b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Moonlight L3 15B 16K?

Moonlight L3 15B 16K is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Moonlight L3 15B 16K by Darkknight535 — VRAM 30GB, 8K context

Name: Moonlight L3 15B 16K
Author: Darkknight535

Moonlight L3 15B 16K is an open-source language model by Darkknight535. Features: 15b LLM, VRAM: 30GB, Context: 8K, LLM Explorer Score: 0.15.

Autotrain compatible Conversational Endpoints compatible Llama Not-for-all-audiences Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/Darkknight535/Moonlight-L3-15B-16k

Moonlight L3 15B 16K Parameters and Internals

Model Type

text generation

Additional Notes

Automatically speaks as other NPCs. Creative and coherent response generation similar to Character.ai.

Training Details

Data Sources:

RP datasheets, dark nsfw datasheets

Methodology:

Fine-tuned using Llama-3-instruct base with context preset. Fine-tuned for 16k context length through rope scaling.

Context Length:

16000

Release Notes

Version:

L3-15B

Notes:

Enhancements include fine-tuning on RP and dark nsfw datasheets. Supports 16k context length through rope scaling.

LLM Name	Moonlight L3 15B 16K
Repository 🤗	https://huggingface.co/Darkknight535/Moonlight-L3-15B-16k
Model Size	15b
Required VRAM	30 GB
Updated	2025-09-23
Maintainer	Darkknight535
Model Type	llama
Model Files	5.0 GB: 1-of-7 5.0 GB: 2-of-7 4.9 GB: 3-of-7 4.9 GB: 4-of-7 5.0 GB: 5-of-7 4.9 GB: 6-of-7 0.3 GB: 7-of-7
Model Architecture	LlamaForCausalLM
Context Length	8192
Model Max Length	8192
Transformers Version	4.45.1
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	<\|end_of_text\|>
Vocabulary Size	128256
Torch Data Type	bfloat16

Rank the Moonlight L3 15B 16K 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

Best Alternatives to Moonlight L3 15B 16K

Best Alternatives	Context / RAM	Downloads	Likes
Nsfw 4dot1 15B Hopefully Right	128K / 30.1 GB	7	0
Moonlight L3 15B V2.5 64K	64K / 30 GB	17	6
Moonlight L3 15B V2 64K	64K / 30 GB	9	8
Hamanasu 15B Instruct	16K / 29.4 GB	473	9
L3 Aethora 15B V2	8K / 30.1 GB	22	42
Meta Llama 3 15B Instruct	8K / 23.9 GB	3	1
OpenCrystal 15B L3 V3	8K / 30 GB	13	10
OpenCrystal L3 15B V2.1	8K / 30 GB	3	5
OpenCrystal 15B L3 V2	8K / 30 GB	3	6
L3 Aethora 15B	8K / 30.1 GB	728	36

Note: green Score (e.g. "73.2") means that the model is better than Darkknight535/Moonlight-L3-15B-16k.