What are the hardware requirements for Synesthesia 3.1 Task Arithmetic?

Synesthesia 3.1 Task Arithmetic requires approximately 16.1 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Synesthesia 3.1 Task Arithmetic and how large is it?

Synesthesia 3.1 Task Arithmetic is developed by bunnycore, a model with 8b 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 Synesthesia 3.1 Task Arithmetic?

Synesthesia 3.1 Task Arithmetic is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Synesthesia 3.1 Task Arithmetic by bunnycore — VRAM 16.1GB, 128K context

Name: Synesthesia 3.1 Task Arithmetic
Author: bunnycore

Synesthesia 3.1 Task Arithmetic is an open-source language model by bunnycore. Features: 8b LLM, VRAM: 16.1GB, Context: 128K, Code Generating, Merged, LLM Explorer Score: 0.14.

Merged Model Arxiv:2212.04089 Base model:bunnycore/hyperllam... Base model:solshine/reflection... Base model:unsloth/meta-llama-... Codegen Endpoints compatible Llama Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/bunnycore/Synesthesia-3.1-task_arithmetic

Synesthesia 3.1 Task Arithmetic Parameters and Internals

Additional Notes

This model is a merge of various pre-trained language models using mergekit.

LLM Name	Synesthesia 3.1 Task Arithmetic
Repository 🤗	https://huggingface.co/bunnycore/Synesthesia-3.1-task_arithmetic
Base Model(s)	Replete-AI/Replete-LLM-V2-Llama-3.1-8b Solshine/reflection-llama-3.1-8B-Solshine-trainround3-16bit unsloth/Meta-Llama-3.1-8B bunnycore/HyperLlama-3.1-8B Replete-AI/Replete-Coder-V2-Llama-3.1-8b Replete-AI/Replete-LLM-V2-Llama-3.1-8b Solshine/reflection-llama-3.1-8B-Solshine-trainround3-16bit unsloth/Meta-Llama-3.1-8B bunnycore/HyperLlama-3.1-8B Replete-AI/Replete-Coder-V2-Llama-3.1-8b
Merged Model	Yes
Model Size	8b
Required VRAM	16.1 GB
Updated	2026-06-28
Maintainer	bunnycore
Model Type	llama
Model Files	5.0 GB: 1-of-4 5.0 GB: 2-of-4 4.9 GB: 3-of-4 1.2 GB: 4-of-4
Generates Code	Yes
Model Architecture	LlamaForCausalLM
Context Length	131072
Model Max Length	131072
Transformers Version	4.44.1
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	<\|finetune_right_pad_id\|>
Vocabulary Size	128256
Torch Data Type	float16

Rank the Synesthesia 3.1 Task Arithmetic 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 Synesthesia 3.1 Task Arithmetic

Best Alternatives	Context / RAM	Downloads	Likes
CoolerCoder 8B Model Stock	128K / 16.1 GB	34	2
Replete Coder V2 Llama 3.1 8B	128K / 16.1 GB	19	9
Seed Coder 8B Reasoning	64K / 32.9 GB	496	154
Seed Coder 8B Reasoning Bf16	64K / 16.5 GB	161	23
Zeta 2.1	32K / 16.5 GB	2053	76
Zeta 2	32K / 16.5 GB	422	187
Seed Coder 8B Instruct	32K / 16.5 GB	4871	118
Seed Coder 8B Base	32K / 16.5 GB	530	74
Seed Coder 8B Instruct V1	32K / 32.9 GB	43	2
Seed Coder 8B Instruct	32K / 16.5 GB	25	1

Note: green Score (e.g. "73.2") means that the model is better than bunnycore/Synesthesia-3.1-task_arithmetic.