AMD Radeon RX 7900 XTX
24 GB · 24GB GDDR6 · single-card node · $800 Card · $2,000 Full rig
Quick decision verdict
AMD Radeon RX 7900 XTX outputs 75.5 tok/s at $0.063/M tokens (rank #8 of 28 measured rigs on unit cost), breaking even in 390 days at full load.
- Generation speed rank #9 of 29 · System power rank #18 (475 W).
- For a lower upfront build cost, compare against 2x RTX 3060 12 GB ($1,814)
- If you need faster interactive token speed, compare against RTX PRO 6000 Blackwell (171.1 tok/s)
How you'll use it
| Fits? | Source | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 7900 XTX 24GB GDDR6 · single-card node | 24 GB | 1,459.3 | 75.5 | Roomy | 475 W | $2,000 | 44.5 | $0.063 | $1.71 | $6.85 | 389 days | Measured |
Selected hardware
AMD Radeon RX 7900 XTX
24 GB · 24GB GDDR6 · single-card node
Read tok/s
1,459.3Write tok/s
75.5Power
475 WOutput/day
44.5Cost/M tok
$0.063Power/day
$1.71Saved/day
$6.85Payback
389 daysFAQ
- How fast does AMD Radeon RX 7900 XTX run local LLMs?
- AMD Radeon RX 7900 XTX (24 GB) measures 75.5 tokens/s generation and 1,459 tokens/s prompt reading on Qwen3.8-Flash-Next (snapshot 2026-10-07).
- Is AMD Radeon RX 7900 XTX worth buying over cloud LLM APIs?
- At $0.063/M tokens local cost, AMD Radeon RX 7900 XTX breaks even in 390 days at 24h/day utilization, or 2,333 days at 4h/day — making daily utilization the main deciding factor.