Builds / AMD Radeon RX 7800 XT 16GB
Radeon RX 7800 XT 16 GB running Qwen3.8 27B IQ3_S (GSQ-RCO, MTP-head GGUF) at 50.2 tok/s
Source: llm-bench.io ↗📏 How measured: reported by the original source; RigBase re-checked the numbers at that source
- GPU(s)
- AMD Radeon RX 7800 XT 16GB
Amazon · eBay (used) · Newegg - VRAM / unified memory (GB)
- 16
- CPU
- not stated
- Motherboard
- not stated
- RAM
- not stated
- PSU
- not stated
- Other
- not stated
- Reported total cost
- not stated
- Power
- not stated
Measured speed (generation tokens/sec)
| Model | Quant | Engine | tok/s | Prompt processing (tok/s) | Details |
|---|---|---|---|---|---|
| Qwen3.8 27B | Qwen3.8-27B-GSQ-RCO-IQ3_S-mtp-ASCII-Condensed (community IQ3_S GGUF with MTP head) | llama.cpp (tool label "q5 q5 -1 budget manual") | 50.2 | 333 (prompt length not stated) | ⚡ spec-decode (MTP)context: 128,000 ctx, thinking used📏 llm-bench.io submission (2026-10-09), memory 16.5 GB; 2nd fastest of 7 runs of this GGUF on this card (median 49.0). The GGUF ships an MTP head and the page does not say whether drafting was on, so it is flagged as speculative and kept out of plain-decode rankings |
Prompt processing = prefill speed (how fast the prompt is read) as reported by the source, with the prompt length it was measured at; blank = the source does not report it. est. TTFT @8k is calculated from the measured prefill speed (8,192 ÷ prefill tok/s), shown only when prefill was measured on a prompt of 512+ tokens. It is not a measurement and ignores load time and caching. measured TTFT is the source's own time-to-first-token. Why prefill matters for coding assistants.
llm-bench.io community submission. Rest of the system not stated.
Source: https://llm-bench.io/benchmarks/cmv0k0izy00md01npde7f66nx
Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.
Rent before you buy: try this model on a cloud GPU for a few dollars first: Runpod (referral link; sign up with Google to get the new-user bonus)