Recipe
qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp1-ga1f0z83
qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp1-ga1f0z83Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- vllm
- Engine version
- 0.27.2rc1.dev77+gac7509e2b XPU
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 131,072
- Max concurrency
- 64
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/Qwen/Qwen3.8-27B- Repository
- Qwen/Qwen3.8-27B
- Status
- known
- Link type
- Exact Hub repository
Public Hugging Face repository confirmed by the Hub API.
Candidate: useful compatibility or speed evidence. The registry does not offer Run until promotion requirements are met.
Observed configuration
vllm
Evidence only · candidate · reference
Candidate evidence — not a Run contract
Observed source tokens
Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.
vllmserve/model--quantizationgptq--dtypefloat16--max-model-len131072--gpu-memory-utilization0.88--kv-cache-dtypefp8--max-num-seqs64--max-num-batched-tokens8192--no-enable-prefix-caching--language-model-only--speculative-config{method:mtp,num_speculative_tokens:4}
| Flag | Value |
|---|---|
--quantization | gptq |
--dtype | float16 |
--max-model-len | 131072 |
--gpu-memory-utilization | 0.88 |
--kv-cache-dtype | fp8 |
--max-num-seqs | 64 |
--max-num-batched-tokens | 8192 |
--speculative-config | {method:mtp,num_speculative_tokens:4} |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 131,072 | 1,774 | 83.7 | — | observed | qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp1-ga1f0z83-sweep |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- hardware count
- 1
- hardware id
- intel-arc-pro-b70-32gb
- id
- qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp1-ga1f0z83
- model instance id
- qwen-qwen3-8-27b--gptq-int4
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- qwen3-8-27b-gptq-int4-intel-arc-pro-b70-32gb-vllm-tp1-ga1f0z83-sweep
- status
- candidate
capabilities
engine
- name
- vllm
- version
- 0.27.2rc1.dev77+gac7509e2b XPU
serving
- max concurrency
- 64
- max context tokens
- 131,072
- tensor parallel
- 1
Provenance & metadata (3)
facts
- capabilities.chat · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
- capabilities.reasoning · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
- capabilities.tools · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
- capabilities.vision · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
- engine.graph mode · provenance · captured at
- 2026-08-30T09:10:02Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
- serving.kv cache tokens · provenance · captured at
- 2026-08-30T09:10:02Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown
- serving.max concurrency · provenance · captured at
- 2026-09-01T10:20:39Z
serving.max concurrency · reason explicit-source-server-capacity
serving.max concurrency · state known
- serving.max context tokens · provenance · captured at
- 2026-09-01T10:20:39Z
serving.max context tokens · reason explicit-source-context-limit
serving.max context tokens · state known
metadata
- localmaxxing · backend
- xpu
- localmaxxing · batch size
- 1
- localmaxxing · hardware label
- Intel Arc Pro B70
- localmaxxing · notes
- Single-stream C1. vLLM XPU nightly vllm/vllm-openai-xpu@sha256:f01e24f6c7ff01f1e0662234255a1372297d1dbd89d003cf13c8fad3eab1ba4f, vllm-xpu-kernels 0.1.12.3. Model SergiioB/Qwen3.8-27B-GPTQ-Int4-sym-G128-MTP-BF16 (Qwen3_5ForConditionalGeneration, dense 27B, GPTQ-INT4 g128 sym desc_act=false, preserved BF16 MTP head). Patches: patch_mtp_nightly.py + patch_mtp_boundary.py. fp8 KV cache, gpu-memory-utilization 0.88, scheduler 8192, max-num-seqs 64, prefix cache OFF (--no-enable-prefix-caching, zero cache-query delta). MTP4 speculative decode, 1-layer draft. tokSOut = client post-first rate at p512/g128, median n=5 (not engine-native decode); tokSPrefill = actual input tokens / client TTFT at p8192/g1, median n=5, no-spec mode (not llama-bench pp). 230 W configured cap; measured campaign-window mean draw 196 W (MTP4 mode). E2 self-reported; independent reproduction pending. Measured 2026-08-15. Claims-audited: 83.7 tok/s MTP4 p512/g128 n=5 range 78.3-84.4; exact-128K full-context p130944/g128 56.3 tok/s n=1. 2000 t/s prefill NOT reached; max cold-input median 1851 t/s (no-spec p2048/g1).
- localmaxxing · observed command
- vllm serve /model --quantization gptq --dtype float16 --max-model-len 131072 --gpu-memory-utilization 0.88 --kv-cache-dtype fp8 --max-num-seqs 64 --max-num-batched-tokens 8192 --no-enable-prefix-caching --language-model-only --speculative-config {"method":"mtp","num_speculative_tokens":4}
- localmaxxing · run id
- cmsur82fz06svms01ga1f0z83
- localmaxxing · tokenized · arguments
- vllm, serve, /model, --quantization, gptq, --dtype, float16, --max-model-len, 131072, --gpu-memory-utilization, 0.88, --kv-cache-dtype, fp8, --max-num-seqs, 64, --max-num-batched-tokens, 8192, --no-enable-prefix-caching, --language-model-only, --speculative-config, {method:mtp,num_speculative_tokens:4}
- localmaxxing · tokenized · fidelity
- faithful
provenance
- captured at
- 2026-08-30T09:10:02Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-30T09:10:02Z | normalized-recipe | www.localmaxxing.com/en/runs/cmsur82fz06svms01ga1f0z83 ↗ |