RUNLOCALAIv38
->Will it run?Best GPUCompareTroubleshootStartLearnPulseModelsHardwareToolsBench
Run check
  1. >
  2. Home
  3. /Hardware
  4. /AMD EPYC 9005 (Zen 5, Turin)
UNIT · AMD · CPU
768 GB UNIFIEDworkstation·Reviewed June 2026

AMD EPYC 9005 (Zen 5, Turin)

AMD · HARDWARE
AMD EPYC 9005 (Zen 5, Turin)

No editorial image yet — generic vendor mark shown. Credentials in spec table below.

AMD's Zen 5 server CPU: up to 192 cores, 12-channel DDR5-6400 (~614 GB/s per socket), supporting terabyte-scale RAM. Anchors the CPU-inference strategy — running the full DeepSeek-R1 671B entirely in system memory.

Released 2024·614 GB/s memory bandwidth
RUNLOCALAI SCORE
See full leaderboard →
161/ 1000
DD-tier
Estimated
Throughput
178/ 500
VRAM-fit
0/ 200
Ecosystem
40/ 200
Efficiency
12/ 100

Sub-scores sum to 230 / 1000. Headline = 230 × 0.70 (Estimated-confidence discount) = 161. This is an algorithmic performance-tier score — distinct from, and often lower than, the editorial “Our verdict” below, which weighs value and real-world fit (especially for hardware we haven’t measured yet). How scoring works →

Extrapolated from 614 GB/s bandwidth — 61.4 tok/s estimated. No measured benchmarks yet.

WORKLOAD FIT
Try other hardware →

Plain-English: Doesn't fit modern chat models usefully — vision models won't fit.

7B chat✗
Doesn't fit
14B chat✗
Doesn't fit
32B chat✗
Doesn't fit
70B chat✗
Doesn't fit
Coding agent✗
Doesn't fit
Vision (≤8B VLM)✗
Doesn't fit
Long context (32K)✗
Doesn't fit
✓Comfortable — fits with headroom
~Tight — works, no slack
△Marginal — needs aggressive quant
✗Doesn't fit usefully

Verdicts extrapolated from catalog VRAM + bandwidth + ecosystem flags. Hover any chip for the rationale. Want measured numbers? Submit your own run with runlocalai-bench --submit.

BLK · VERDICT

Our verdict

OP · Fredoline Eruo|VERIFIED JUN 18, 2026
7.7/10

What it is

EPYC 9005 'Turin' is AMD's Zen 5 server platform: up to 192 cores, 12-channel DDR5-6400 for ~614 GB/s of memory bandwidth per socket, and support for terabyte-plus RAM capacities. It's the anchor of the catalog's CPU-inference category (previously empty).

Relevance to local AI

CPU inference is a real, distinct local-AI strategy that GPUs can't match at the extreme high end: with 768GB-1.5TB of system RAM, an EPYC 9005 box runs the full DeepSeek-R1 671B (and other giant MoE models) entirely in memory — something no single GPU and even few unified-memory machines can do. The tradeoff is speed: token generation is bandwidth-bound and slow (single-digit to low-double-digit tok/s on the largest models), but for running models that simply don't fit anywhere else, or for batch/offline workloads, a big-RAM EPYC is the answer. MoE models especially benefit since only a fraction of params activate per token.

Bottom line

The go-to for fitting the absolute largest models (671B-class) locally via system RAM, where capacity beats speed. An enthusiast/datacenter pick — slow per-token but uniquely capable of models nothing else can hold.

BLK · OVERVIEW

Overview

AMD's Zen 5 server CPU: up to 192 cores, 12-channel DDR5-6400 (~614 GB/s per socket), supporting terabyte-scale RAM. Anchors the CPU-inference strategy — running the full DeepSeek-R1 671B entirely in system memory.

Retailers we'd check:Amazon

Search-fallback link — editorial hasn't yet curated a retailer URL for this card.

Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.

BLK · SPECS

Specs

System RAM (typical)768 GB
Power draw (peak)400 W
Released2024
Backends

Models that fit

Open-weight models small enough to run on AMD EPYC 9005 (Zen 5, Turin) with usable context.

all-MiniLM-L6-v2
0.022B · other
FLUX.1 [dev]
12B · other
Qwen 3.5 235B-A17B (MoE)
397B · qwen
Qwen 3 235B-A22B
235B · qwen
Qwen 3 0.6B
0.6B · qwen
BGE Large EN v1.5
0.335B · other
Llama 4 Scout
109B · llama
DeepSeek V4 Flash (284B MoE)
284B · deepseek

Frequently asked

Does AMD EPYC 9005 (Zen 5, Turin) support CUDA?

AMD EPYC 9005 (Zen 5, Turin) does not support CUDA. Use Vulkan-compatible tools (llama.cpp Vulkan backend) or check vendor-specific runtimes.

Where next?

Buyer guides
  • Best GPU for local AI →
  • Best laptop for local AI →
  • Best Mac for local AI →
  • Best used GPU for local AI →
Troubleshooting
  • CUDA out of memory →
  • Ollama running slowly →
  • ROCm not detected →
  • Model keeps crashing →

Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.

RUNLOCALAI

Independently operated catalog for local-AI hardware and software. Hand-written verdicts. Source-cited claims. Reproducible commands when we have them.

OP·Fredoline Eruo
DIR
  • Models
  • Hardware
  • Tools
  • Benchmarks
TOOLS
  • Will it run?
  • Compare hardware
  • Cost vs cloud
  • Choose my GPU
  • Prompting kits
  • Quick answers
REF
  • All buyer guides
  • Learn local AI
  • Methodology
  • Glossary
  • Errors KB
  • Trust
EDITOR
  • About
  • Author
  • How we make money
  • Editorial policy
  • Contact
LEGAL
  • Privacy
  • Terms
  • Sitemap
MAIL · MONTHLY DIGEST
Get monthly local AI changes
Monthly recap. No spam.
DISCLOSURE

Some links on this site are affiliate links (Amazon Associates and other first-class retailers). When you buy through them, we earn a small commission at no extra cost to you. Affiliate links do not influence our verdicts — there are cards we rate highly that we don't have affiliate relationships with, and cards that sell well that we refuse to recommend. Read more →

© 2026 runlocalai.coIndependently operated
RUNLOCALAI · v38
Compare alternatives

Hardware worth comparing

The closest alternatives by price, memory bandwidth, and form factor, plus a step up and down — so you can frame the buying decision against real options.

Closest matches
Similar price, bandwidth & form factor
  • NVIDIA RTX 5000 Ada Generation
    nvidia · 32 GB VRAM
    9.5/10
  • NVIDIA RTX PRO 4000 Blackwell
    nvidia · 24 GB VRAM
    7.3/10
  • Intel Arc Pro B60 24GB
    intel · 24 GB VRAM
    7.6/10
  • NVIDIA L40S
    nvidia · 48 GB VRAM
    10.0/10
  • NVIDIA L40
    nvidia · 48 GB VRAM
    10.0/10
  • NVIDIA RTX PRO 4500 Blackwell
    nvidia · 32 GB VRAM
    7.5/10
Step up
More capable — more memory or a higher tier
No verdicted hardware in the next tier up yet.
Step down
Lighter — cheaper or more constrained
  • NVIDIA GeForce RTX 4090 Mobile
    nvidia · 16 GB VRAM
    7.3/10
  • NVIDIA GeForce RTX 5070
    nvidia · 12 GB VRAM
    7.6/10
  • NVIDIA GeForce RTX 4070 Ti Super
    nvidia · 16 GB VRAM
    8.1/10