Speeds are as each source reported them. GPU count, offloading and concurrent requests change them, and prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing two reports.
- generation:
- 15.6 tokens/s
- prompt processing (prefill):
- 339.4 tokens/s
- quant:
- Q4_K_M (GGUF)
User compares Qwen3.8-27B Q4_K_M on an Intel Arc Pro B60 24GB under WSL2 with Mesa Dozen against the native Windows Vulkan build of llama.cpp b11889. Native Windows gives 339.4 t/s prompt and 15.6 t/s generation with -ngl 99; the WSL2 run segfaults during generation after 49.9 t/s prompt.
The crash is in libLLVM-14.so from Intel's Windows driver package, at the same offset across four crashes, two quants and both llama-bench and llama-cli. It runs at -ngl 0 or 1 and crashes from -ngl 2 up, with 23.6 GiB VRAM free.
Other models on the same card are 8 to 40 times slower under WSL2: Ornith 1.5 9B Q8_0 at 141.7 / 4.1 versus 975.8 / 36.2, and Qwen3-Coder-30B-A3B Q4_K_M at 34 / 2.5 versus 788.8 / 99.2. Dozen reports matrix cores none and a non-conformant Vulkan warning, while native Windows reports KHR_coopmat.
- generation:
- 53.0 tokens/s
- kv:
- f16
coding
User reports Qwen3.8-27B at 53 tok/s mean decode on an Intel Arc Pro B60 24GB, with a 64K context window.
Setup is llama.cpp on the Vulkan backend with f16 KV cache, -ub 256 and a q8 draft KV cache, using DFlash2 speculative decoding and Intel's FA prefill kernel. Driver 32.0.101.8976 on Windows 11.
The user notes smaller cards (B580/B570/B50/A770) will not fit this 27B, though the Vulkan tuning applies. The same card also runs LTX-2.5 video generation via ComfyUI on XPU.
- generation:
- 53.3 tokens/s
- prompt processing (prefill):
- 420.0 tokens/s
- quant:
- Q4_K_M (GGUF)
- kv:
- F16
codingagentic
User reports Qwen3.8-27B at 53.3 tok/s per-request decode (47 to 58 across eight requests) on an Intel Arc Pro B60 24GB.
Setup is llama.cpp Vulkan build b11524 with Q4_K_M and an F16 KV cache, a DFlash2 draft model, one request at a time, 32K context, the whole model on the card.
The repo's 42.3 tok/s headline is wall-clock eval-suite throughput with prompt processing included, not a decode rate; stock llama.cpp with no drafter gave 16 tok/s in llama-bench tg32. Prompt reading at 32K was 420 tok/s with the Intel FA kernel, versus 118 tok/s stock.
- generation:
- 22.0 tokens/s
- quant:
- INT4
coding
User reports Qwen3.8-27B at a stable 22 t/s generation with up to 95k context on an Intel Arc Pro B60 24GB connected via Oculink to a GMKTEC M7 Pro with 32 GB RAM.
Setup is intel-llm-scaler with an INT4 quant. The user also tried a Swift-Qwen3.8 fine-tune and MTP, but MTP limited context to about 50k and produced garbage quickly. llama.cpp was slower than vLLM on both Vulkan and SYCL.
User asks whether better options exist for a 24 GB card than Qwen3.8-27B, mentioning Qwen3.8-Next-Flash as likely too heavy.
- generation:
- 91.9 tokens/s
- prompt processing (prefill):
- 1760.0 tokens/s
- quant:
- Q4_0 (GGUF)
agenticcoding
User reports Nemotron 3.5 Lightning 30B-A3B at 91.91 tok/s decode on 2x Intel Arc Pro B60 24GB.
Setup is llama.cpp SYCL with Q4_0 weights and an MTP Q8_0 drafter at --spec-draft-n-max 7, 22.18 GiB VRAM, 1,760 tok/s prefill at 12K context.
MTP acceptance is 99.5-100% at every n-max; the model needs the whole card with no co-residence.
- generation:
- 15.3 tokens/s
- prompt processing (prefill):
- 513.4 tokens/s
- quant:
- Q4_K_M (GGUF)
- kv:
- Q8_0
User benchmarks Qwen3.8-27B at 15.34 t/s generation and 513.39 t/s prompt processing on a single Intel Arc Pro B60.
Setup is llama.cpp build b10452 with Vulkan backend, Q4_K_M GGUF weights and Q8_0 KV cache, full GPU offload with Flash Attention, at 512 tokens of context.
A second run on two Arc Pro B60 cards gives 14.99 t/s decode and 507.93 t/s prefill at 512 tokens; dual-card prefill scales to 643.16 t/s at 1k, 718.37 t/s at 2k and 149.30 t/s at 64k, while decode stays near 15 t/s.
- generation:
- 24.0 tokens/s
- prompt processing (prefill):
- 609.5 tokens/s
- quant:
- Q4_K_M (GGUF)
- kv:
- Q8_0
User reports Qwen3.8-27B at 24.00 tok/s decode on dual Intel Arc Pro B60 24GB (48GB total).
Setup is llama.cpp build b11100 with SYCL F16 JIT, Q4_K_M GGUF, Q8_0 KV cache, 131072 context, and native embedded Q8_0 MTP draft speculative decoding.
Prefill is 609.47 tok/s at 512 tokens, scaling to 925.39 tok/s at 4k and 570.24 tok/s at 128k. Stock non-MTP decode is 16.13 tok/s; live llama-server chat with MTP ranges 20.00-30.29 tok/s.
- generation:
- 8.3 tokens/s
- prompt processing (prefill):
- 112.6 tokens/s
- quant:
- Q4_K_XL (GGUF)
User benchmarks Qwen3.5 35B-A3B at 8.34 t/s generation and 112.62 t/s prompt processing on an Intel Arc Pro B60 24GB.
Setup is llama.cpp build 8175 with the SYCL backend, Q4_K_XL GGUF weights, full GPU offload (-ngl 100).
User reports the card is ok for chat-length context on smaller models but slow for agentic tools like opencode/claude code, where initial prompt response can take 5-10 minutes. A comparison run of the same model on an RTX Pro 4500 Blackwell reached 133.47 t/s generation and 3807.62 t/s prompt processing.
- generation:
- 22.2 tokens/s
- prompt processing (prefill):
- 245.5 tokens/s
- quant:
- Q4_K (GGUF)
codingmathlong-context
User benchmarks a Qwen3.8-27B Ridge Intel Arc tuned GGUF on an Intel Arc Pro B60 24GB, reporting 22.23 tok/s pure autoregressive decode and 245.54 tok/s prompt prefill.
Setup is llama.cpp with Q4_K weights, -ngl 99 and -fa on, on a single 24GB card.
With embedded MTP speculative decoding the tuned Ridge build reaches 41.28 tok/s at 93.4% acceptance, and the DAS Lab IQ3_S tuned build reaches 41.88 tok/s at 92.9% acceptance. Stock IQ3_XXS decodes at 8.10 tok/s and stock Unsloth UD-Q4_K_S at 12.97 tok/s. At ~150k-token depth decode falls to 7.91 tok/s with DFlash2 and 7.21 tok/s with native MTP.