Results availableIntel Arc Pro B70

Intel Arc Pro B70 benchmarks

Explore models and exact run configurations. Measurement profiles, resources and evidence stay attached to each run.

Model cards
37
Detail pages
56
Engines
3

Models with detail configurations

A model card remains stable while its available run configurations can grow over time.

37 MODELS

AutoModelForCausalLM 0.5B

Qwen2.5-0.5B-Instruct

Qwen2.5-0.5B-Instruct benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

Qwen2ForCausalLM 0.6B

Qwen3-0.6B-BF16

Qwen3-0.6B-BF16 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 0.6B

Qwen3-0.6B-FP8

Qwen3-0.6B-FP8 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

Qwen2ForCausalLM 0.8B

Qwen3.5-0.8B-BF16

Qwen3.5-0.8B-BF16 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 1.5B

DeepSeek-R1-Distill-Qwen-1.5B

DeepSeek-R1-Distill-Qwen-1.5B benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

Qwen2ForCausalLM 2B

Qwen3.5-2B-BF16

Qwen3.5-2B-BF16 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 3B

Qwen2.5-3B-Instruct-AWQ

Qwen2.5-3B-Instruct-AWQ benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 3B

Qwen2.5-3B-Instruct

Qwen2.5-3B-Instruct benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

Qwen2ForCausalLM 4B

Qwen3.5-4B-BF16

Qwen3.5-4B-BF16 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 7B

Qwen2.5-7B-Instruct

Qwen2.5-7B-Instruct benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 7B

Qwen2.5-VL-7B-Instruct

Qwen2.5-VL-7B-Instruct benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 7B

Qwen2.5-7B-Instruct-AWQ

Qwen2.5-7B-Instruct-AWQ benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 7B

Qwen2.5-Coder-7B-Instruct

Qwen2.5-Coder-7B-Instruct benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

Qwen2ForCausalLM 8B

Qwen3-8B-BF16

Qwen3-8B-BF16 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 8B

Qwen3-VL-8B-Instruct

Qwen3-VL-8B-Instruct benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 8B

Qwen3-8B-AWQ

Qwen3-8B-AWQ benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

Qwen2ForCausalLM 9B

Qwen3.5-9B-BF16

Qwen3.5-9B-BF16 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 14B

Qwen2.5-14B-Instruct-AWQ

Qwen2.5-14B-Instruct-AWQ benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 30B

Qwen3-30B-A3B-GPTQ-Int4

Qwen3-30B-A3B-GPTQ-Int4 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM unknown

phi-4

phi-4 benchmarked on Intel Arc Pro B70 (32GB) with SGLang-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 14B

DeepSeek-R1-Distill-Qwen-14B

DeepSeek-R1-Distill-Qwen-14B benchmarked on Intel Arc Pro B70 (32GB) with SGLang-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 14B

Qwen2.5-14B-Instruct

Qwen2.5-14B-Instruct benchmarked on Intel Arc Pro B70 (32GB) with SGLang-XPU (TP=1).

License: Unavailable

AutoModelForCausalLM 0.6B

Qwen3-0.6B-GGUF

Qwen3-0.6B-GGUF benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-1).

License: Unavailable

Available run configurations Exact measurement
TP=1 · No detail page No publication-eligible measurement exists. Unavailable
TP=2 · No detail page No publication-eligible measurement exists. Unavailable
TP=4 · No detail page No publication-eligible measurement exists. Unavailable
Layer-Split-1 · Intel Arc Pro B70 (32GB) llama.cpp-SYCL · BF16 · Exclusive use C=1 · 337.9 tok/s Peak 2.49 GiB

AutoModelForCausalLM 0.5B

Qwen2.5-0.5B-Instruct-GGUF

Qwen2.5-0.5B-Instruct-GGUF benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-1).

License: Unavailable

Available run configurations Exact measurement
TP=1 · No detail page No publication-eligible measurement exists. Unavailable
TP=2 · No detail page No publication-eligible measurement exists. Unavailable
TP=4 · No detail page No publication-eligible measurement exists. Unavailable
Layer-Split-1 · Intel Arc Pro B70 (32GB) llama.cpp-SYCL · BF16 · Exclusive use C=1 · 205.39 tok/s Peak 0.95 GiB

AutoModelForCausalLM 0.8B

Qwen_Qwen3.5-0.8B-GGUF

Qwen_Qwen3.5-0.8B-GGUF benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-1).

License: Unavailable

Available run configurations Exact measurement
TP=1 · No detail page No publication-eligible measurement exists. Unavailable
TP=2 · No detail page No publication-eligible measurement exists. Unavailable
TP=4 · No detail page No publication-eligible measurement exists. Unavailable
Layer-Split-1 · Intel Arc Pro B70 (32GB) llama.cpp-SYCL · BF16 · Exclusive use C=1 · 293.68 tok/s Peak 1.43 GiB

AutoModelForCausalLM 3B

Qwen2.5-3B-Instruct-GGUF

Qwen2.5-3B-Instruct-GGUF benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-1).

License: Unavailable

Available run configurations Exact measurement
TP=1 · No detail page No publication-eligible measurement exists. Unavailable
TP=2 · No detail page No publication-eligible measurement exists. Unavailable
TP=4 · No detail page No publication-eligible measurement exists. Unavailable
Layer-Split-1 · Intel Arc Pro B70 (32GB) llama.cpp-SYCL · BF16 · Exclusive use C=1 · 153.12 tok/s Peak 2.93 GiB

AutoModelForCausalLM 30B

Qwen3-30B-A3B-GGUF

Qwen3-30B-A3B-GGUF benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-1).

License: Unavailable

AutoModelForCausalLM 35B

Qwen3.6-35B-A3B-FP8

Qwen3.6-35B-A3B-FP8 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=2).

License: Unavailable

AutoModelForCausalLM 27B

Qwen3.8-27B-FP8

Qwen3.8-27B-FP8 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=2).

License: Unavailable

AutoModelForCausalLM 26B

gemma-4-26B-A4B-it-FP8-Dynamic

gemma-4-26B-A4B-it-FP8-Dynamic benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=2).

License: Unavailable

AutoModelForCausalLM 31B

gemma-4-31B-it-FP8-Dynamic

gemma-4-31B-it-FP8-Dynamic benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=2).

License: Unavailable

Qwen2ForCausalLM 27B

Qwen3.6-27B-int4-AutoRound

Qwen3.6-27B-int4-AutoRound benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=2).

License: Unavailable

AutoModelForCausalLM 27B

Qwen3.8-27B-INT4-W4A16-AutoRound

Qwen3.8-27B-INT4-W4A16-AutoRound benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=2).

License: Unavailable

GemmaForCausalLM 31B

Gemma-4-31B-it-AutoRound-Int4

Gemma-4-31B-it-AutoRound-Int4 benchmarked on Intel Arc Pro B70 (32GB) with vLLM-XPU (TP=2).

License: Unavailable

AutoModelForCausalLM 42B

Phi-3.5-MoE-instruct-GGUF

Phi-3.5-MoE-instruct-GGUF benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-2).

License: Unavailable

Available run configurations Exact measurement
TP=1 · No detail page No publication-eligible measurement exists. Unavailable
TP=2 · No detail page No publication-eligible measurement exists. Unavailable
TP=4 · No detail page No publication-eligible measurement exists. Unavailable
Layer-Split-2 · Intel Arc Pro B70 (32GB) llama.cpp-SYCL · GGUF · Exclusive use C=1 · 62.28 tok/s Peak 25.25 GiB

AutoModelForCausalLM 26B

gemma-4-26B-A4B-it-qat-q4_0-gguf

gemma-4-26B-A4B-it-qat-q4_0-gguf benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-2).

License: Unavailable

Available run configurations Exact measurement
TP=1 · No detail page No publication-eligible measurement exists. Unavailable
TP=2 · No detail page No publication-eligible measurement exists. Unavailable
TP=4 · No detail page No publication-eligible measurement exists. Unavailable
Layer-Split-2 · Intel Arc Pro B70 (32GB) llama.cpp-SYCL · GGUF · Exclusive use C=1 · 62.6 tok/s Peak 15.56 GiB

AutoModelForCausalLM 22B

Codestral-22B-v0.1-GGUF

Codestral-22B-v0.1-GGUF benchmarked on Intel Arc Pro B70 (32GB) with llama.cpp-SYCL (Layer-Split-2).

License: Unavailable

Available run configurations Exact measurement
TP=1 · No detail page No publication-eligible measurement exists. Unavailable
TP=2 · No detail page No publication-eligible measurement exists. Unavailable
TP=4 · No detail page No publication-eligible measurement exists. Unavailable
Layer-Split-2 · Intel Arc Pro B70 (32GB) llama.cpp-SYCL · GGUF · Exclusive use C=1 · 34.19 tok/s Peak 15.14 GiB

Filter benchmarks

Available topology
Concurrency · throughput Measured workload only

What accompanies a measurement

WorkloadInput and output lengths, concurrency and cache state belong to the measurement.
RuntimeModel revision, engine and topology describe the exact run.
Separate validationArtifact and validation report must match the same SHA-256 digest. Missing values remain not measured.