What are the hardware requirements for AnFeng V3 Avocet?

AnFeng V3 Avocet requires approximately 69.5 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 AnFeng V3 Avocet and how large is it?

AnFeng V3 Avocet is developed by NLPark, a model with 35b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does AnFeng V3 Avocet perform on standard benchmarks?

AnFeng V3 Avocet has the following published scores: MMLU 74.79. Compare against reference models on this page or on the LLM Explorer leaderboards.

AnFeng V3 Avocet by NLPark — VRAM 69.5GB, 8K context

Name: AnFeng V3 Avocet
Author: NLPark

AnFeng V3 Avocet is an open-source language model by NLPark. Features: 35b LLM, VRAM: 69.5GB, Context: 8K, License: cc-by-nc-nd-4.0, LLM Explorer Score: 0.16, Arc: 67.8, HellaSwag: 87.4, MMLU: 74.8, GSM8K: 74.

Cohere Conversational En Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/NLPark/AnFeng_v3_Avocet

AnFeng V3 Avocet Parameters and Internals

Model Type

text generation

Additional Notes

Sizes are labelled as: Nano, Leap, Pattern, Avocet, Robin, Kestrel.

Supported Languages

Chinese (advanced), English (advanced)

Release Notes

Version:

v3 early preview

Notes:

Released as an early preview for v3 series covering 'Shi-Ci', 'AnFeng', and 'Cecilia' LLM products.

LLM Name	AnFeng V3 Avocet
Repository 🤗	https://huggingface.co/NLPark/AnFeng_v3_Avocet
Model Size	35b
Required VRAM	69.5 GB
Updated	2026-07-05
Maintainer	NLPark
Model Type	cohere
Model Files	4.7 GB: 1-of-15 4.9 GB: 2-of-15 4.9 GB: 3-of-15 4.9 GB: 4-of-15 4.9 GB: 5-of-15 4.9 GB: 6-of-15 4.9 GB: 7-of-15 4.9 GB: 8-of-15 4.9 GB: 9-of-15 4.9 GB: 10-of-15 4.9 GB: 11-of-15 4.9 GB: 12-of-15 4.9 GB: 13-of-15 4.9 GB: 14-of-15 1.1 GB: 15-of-15
Supported Languages	en
Model Architecture	CohereForCausalLM
License	cc-by-nc-nd-4.0
Context Length	8192
Model Max Length	8192
Transformers Version	4.38.2
Tokenizer Class	LlamaTokenizer
Padding Token	<PAD>
Vocabulary Size	256000
Torch Data Type	bfloat16

Rank the AnFeng V3 Avocet 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 AnFeng V3 Avocet

Best Alternatives	Context / RAM	Downloads	Likes
Command R 35B Writer V3	8K / 69.5 GB	7	3
Creative Writer 35B Preview	8K / 69.5 GB	16	0
Frejya 0.8	8K / 69.5 GB	12	0
...eative Writer V0.1 Charlie 35B	8K / 69.5 GB	33	0
Creative Writer V0.1 Alfa 35B	8K / 69.5 GB	28	4
Creative Writer V0.1 Bravo 35B	8K / 69.5 GB	38	0
Aya 23 35B	8K / 69.5 GB	446	291
35B Beta2ep	8K / 69.5 GB	20	19
...ommand R V01 Japanese Instruct	8K / 69.5 GB	58	4
35B Beta Long	8K / 69.5 GB	24	69

Note: green Score (e.g. "73.2") means that the model is better than NLPark/AnFeng_v3_Avocet.