Moore Threads publishes a product page for this card that contains no specification table. Its capacity, bandwidth, interconnect and FP8 figures all come from the company's first public disclosure of the parameters in February 2026, carried by the Chinese technology press, which is why the row is marked vendor claimed rather than official. The vendor page's performance claims are all relative to unnamed competitors, and the three baselines it footnotes, 989 TFLOPS dense FP16 with 3.35 TB/s, 148 TFLOPS with 4.0 TB/s and 312 TFLOPS with 2.0 TB/s, describe an NVIDIA H100, H20 and A100 rather than this card. None of those figures is stored here. Moore Threads states that the part runs every precision from FP8 to FP64 but publishes a rate for FP8 only, so the other rows show support without a figure. No core count, tensor core count, memory type, process node or board power has been published by anyone, and the 8,192 shading cores and 512 tensor cores quoted in some aggregator listings appear in no vendor or first-party document, so they are not recorded.

Moore Threads MTT S5000

Architecture: MUSA 4th GenForm factor: OAMReleased: 2025Spec confidence: Vendor claimed

FP8 (dense)

1,000

TFLOPS

Memory

80 GB

Bandwidth

1.6 TB/s

memory

Rent this GPU Download datasheet

Performance

Peak theoretical throughput by precision type

Precision	Peak
FP64 64-bit floating point	No verified data available
FP32 32-bit floating point	No verified data available
TF32	No verified data available
BF16	No verified data available
FP16 16-bit floating point	No verified data available
FP8 8-bit floating point	1,000TFLOPS
FP6	No verified data available
FP4	No verified data available
INT8	No verified data available

Full Specifications

Memory
Memory	80 GB
Memory Type	No verified data available
Bandwidth	1.6 TB/s
Interface Width	No verified data available
Interconnect & I/O
GPU-to-GPU	MTLink
Interconnect Bandwidth	784 GB/s direction and scope not stated by vendor
Power & Thermal
TDP	No verified data available
Max power	No verified data available
Enterprise Features
Compute APIs	MUSA, Torch-MUSA (PyTorch), PaddlePaddle, Megatron-LM, vLLM, SGLang
General
Form Factor	OAM
Architecture	MUSA 4th Gen
Process Node	No verified data available
Launch Year	2025

Systems Using This GPU

Pre-configured systems featuring the Moore Threads MTT S5000

System	GPU Count	Peak Performance	Total Power	Interconnect
Moore Threads MTT SGX5000 SGX · rack ·	8x MTT S5000	No verified data available	11.4 kW	Eight single-port 400Gb InfiniBand HDR or Ethernet CX7 adapters, one per accelerator, for scale-out; storage on a dual-port 100Gb RoCE adapter and management on gigabit Ethernet. Moore Threads publishes no scale-up fabric bandwidth for the chassis.	View System

Frequently Asked Questions

How many TFLOPS does the Moore Threads MTT S5000 have?

The Moore Threads MTT S5000 delivers 1,000 TFLOPS FP8 at peak. Flopper does not currently have verified FP32 and FP16 throughput figures for it.

What is the power consumption of the Moore Threads MTT S5000?

Flopper does not currently have a verified TDP (Thermal Design Power) figure for the Moore Threads MTT S5000.

How much memory does the Moore Threads MTT S5000 have?

The Moore Threads MTT S5000 is equipped with 80 GB of memory with 1,600 GB/s of memory bandwidth.

What architecture is the Moore Threads MTT S5000 based on?

The Moore Threads MTT S5000 is based on the MUSA 4th Gen architecture, launched in 2025.

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs

Moore Threads MTT S5000

Overview

Performance

Specifications

Full Specifications

Datasheet & Resources

Data Provenance

Primary Source

Data Quality

Systems Using This GPU

Frequently Asked Questions

Stay Updated on GPU Releases

Stay Updated on GPU Specs