Building a 4x R9700 setup for a 10 person startup

Just wanted to share a build I am doing for a client.

$18k.

Specs:

Threadripper 9970x

128gb DDR5 ECC 5600

4x AMD Radeon R9700 AI Pro 32gb(128gb total VRAM)

1600W PSU(GPUs are undervolted and under 210w each)

Engine: using a fork of Radiance to serve Qwen 3.8 Next flash/27b and DSV4 Flash

Results:

16 concurrent sessions

Qwen 3.8 27b MXFP4

6.3-6.8k aggregate prefill tok/s
900 t/s - 80t/s aggregate decode(range: 4k - 128k context each)

Qwen 3.8 Next flash

48K/user | 6,880 tok/s | 538.5 tok/s aggregate prefill/decode

128k/user | 3,050 tok/s | 365 tok/s aggreate prefill/decode

BF16 kv for all.