SAVRN
Search Contact SAVRN

Open-weight model

moe-ft-018a-s0804

by crazyApe crazyape777/moe-ft-018a-s0804

moe-ft-018a-s0804 is an open-weight model from crazyApe. It has 34.7B parameters and a 262,144-token context. At 16-bit it needs about 83.2 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.

Parameters34.7B
Context262,144
Weights69.3 GB
License
AccessOpen weights
Monthly Downloads

Runs On

What it takes to serve moe-ft-018a-s0804 (34.7B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.

PrecisionWeightsMemory neededCheapest setupPer hourAlso fits
16-bit 69.3 GB 83.2 GB 1x MI300X (192 GB)
Vultr
$1.85 1x MI325X $2.00 · 1x MI355X $2.59
8-bit 34.7 GB 41.6 GB 1x MI300X (192 GB)
Vultr
$1.85 1x H100 $1.99 · 1x MI325X $2.00
4-bit 17.3 GB 20.8 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 Sep 20, 2026.

moe-ft-018a-s0804 on every accelerator the SAVRN Index prices, at every precision

Model Card

The publisher has not written a card for this model.

Configuration

Architecture
Qwen3_5MoeForCausalLM
Context length (tokens)
262,144
Layers
40
Hidden size
2,048
Attention heads
16
Key/value heads
2
Head dimension
256
Vocabulary size
248,320
Experts
256
Experts active per token
8
Model type
qwen3_5_moe_text

Identity and Version

Repository
crazyape777/moe-ft-018a-s0804
Publisher
crazyApe
Task
Not stated by the source
Modality
Other
Library
Not stated by the source
Parameters
34.7B parameters
Languages
Not stated by the source
Revision
91e415fae9a63a302a93a7550ef8b653660a8f1e
First published
2026-09-20
Last updated
2026-09-20

Files and Weights

25 files, 69.3 GB in total. The weights are 18 files totalling 69.3 GB in safetensors.

Weights18 files · 69.3 GB
Configuration3 files · 61.8 KB
Tokenizer2 files · 20.0 MB
Other1 file · 7.8 KB
Repository1 file · 1.6 KB
Every file
FileTypeSizeSHA-256
model-00001-of-00018.safetensorsWeights4.1 GB d7358fc4bd98
model-00002-of-00018.safetensorsWeights4.2 GB ae55977d9543
model-00003-of-00018.safetensorsWeights2.1 GB 289736fb7448
model-00004-of-00018.safetensorsWeights4.3 GB 18e30034a988
model-00005-of-00018.safetensorsWeights4.3 GB 37627ecb1e88
model-00006-of-00018.safetensorsWeights3.2 GB 0d9c97cb02a6
model-00007-of-00018.safetensorsWeights4.2 GB d4297f3a29fd
model-00008-of-00018.safetensorsWeights4.3 GB 49b33e763468
model-00009-of-00018.safetensorsWeights3.2 GB bfcb786040bd
model-00010-of-00018.safetensorsWeights4.2 GB b640ea0963e2
model-00011-of-00018.safetensorsWeights4.3 GB cf6a62d146b5
model-00012-of-00018.safetensorsWeights3.2 GB d3bf45b2d310
model-00013-of-00018.safetensorsWeights4.2 GB 1642d09fbb0c
model-00014-of-00018.safetensorsWeights4.3 GB 11fabab9ee5b
model-00015-of-00018.safetensorsWeights3.2 GB 6e402e2505fc
model-00016-of-00018.safetensorsWeights4.2 GB 9af134e61cf5
model-00017-of-00018.safetensorsWeights4.3 GB 4b16a49d1eb5
model-00018-of-00018.safetensorsWeights3.2 GB d77ee54bfc56
config.jsonConfiguration2.3 KB
generation_config.jsonConfiguration213 B
model.safetensors.index.jsonConfiguration59.3 KB
chat_template.jinjaOther7.8 KB
.gitattributesRepository1.6 KB
tokenizer.jsonTokenizer20.0 MB 06b9509352d2
tokenizer_config.jsonTokenizer1.2 KB

License and Download

License
Not stated by the source
Access
Open weights, no gate
Download size
69.3 GB
Download from crazyApe

Released by crazyApe through its official repository on Hugging Face.

Memory Requirements

PrecisionWeights in memory
As published69.3 GB
16-bit69.3 GB
8-bit34.7 GB
4-bit17.3 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Questions About moe-ft-018a-s0804

How much GPU memory does moe-ft-018a-s0804 need?

About 83.2 GB at 16-bit and 20.8 GB at 4-bit: the weights (34.7B parameters) plus a working margin. A long context needs more.

What is the cheapest GPU to run moe-ft-018a-s0804 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 moe-ft-018a-s0804's context length?

262,144 tokens, from the maximum position embeddings in its published configuration.