| The Intel® Core™ 7 150U RPL-U can run AI Inferencing for DeepSeek R1-Distill-Qwen-1.5 B up to 25 tokens/s on iGPU with Batch size 1 at INT4 precision with a TTFT (Time to First Token Latency) of 1493ms with a Power Efficiency of 2.19 Tokens/s/W. | Processor: Intel Core 7 150U TDP=15W; tested in LENOVO ThinkEdge SE30n; Memory: 32GB (1x32GB DDR5 5600MT/s [5200MT/s]), BIOS M6OKT02A, microcode 0x4129, 1x Unknown NIC, 1x Ethernet Connection, 1x Ethernet Controller I226-V, 1x Raptor Lake PCH CNVi WiFi, 1x 238.5G SAMSUNG MZVL8256HFJD-00BLL, Ubuntu 24.04.4 LTS, 6.17.0-14-generic. | As measured by OpenVINO GenAI LLM Bench 2025.4.1 | As of February 2026 |
| The Intel® Core™ 7 150U RPL-U can run AI Inferencing for Phi-4-mini-reasoning up to 12 tokens/s on iGPU with Batch size 1 at INT4 precision with a TTFT (Time to First Token Latency) of 3406ms with a Power Efficiency of 0.94 Tokens/s/W. | Processor: Intel Core 7 150U TDP=15W; tested in LENOVO ThinkEdge SE30n; Memory: 32GB (1x32GB DDR5 5600MT/s [5200MT/s]), BIOS M6OKT02A, microcode 0x4129, 1x Unknown NIC, 1x Ethernet Connection, 1x Ethernet Controller I226-V, 1x Raptor Lake PCH CNVi WiFi, 1x 238.5G SAMSUNG MZVL8256HFJD-00BLL, Ubuntu 24.04.4 LTS, 6.17.0-14-generic. | As measured by OpenVINO GenAI LLM Bench 2025.4.1 | As of February 2026 |
| The Intel® Core™ 7 150U RPL-U can run AI Inferencing for Qwen3-8B up to 7 tokens/s on iGPU with Batch size 1 at INT4 precision with a TTFT (Time to First Token Latency) of 7249ms with a Power Efficiency of 0.41 Tokens/s/W. | Processor: Intel Core 7 150U TDP=15W; tested in LENOVO ThinkEdge SE30n; Memory: 32GB (1x32GB DDR5 5600MT/s [5200MT/s]), BIOS M6OKT02A, microcode 0x4129, 1x Unknown NIC, 1x Ethernet Connection, 1x Ethernet Controller I226-V, 1x Raptor Lake PCH CNVi WiFi, 1x 238.5G SAMSUNG MZVL8256HFJD-00BLL, Ubuntu 24.04.4 LTS, 6.17.0-14-generic. | As measured by OpenVINO GenAI LLM Bench 2025.4.1 | As of February 2026 |
| The Intel® Core™ 7 150U (RPL-U) can support up to 3 streams on CPU, 8 streams on iGPU @14.95fps target FPS with batch size 1 | Processor: Intel Core 7 150U TDP=15W; tested in LENOVO ThinkEdge SE30n; Memory: 32GB (1x32GB DDR5 5600MT/s [5200MT/s]), BIOS M6OKT02A, microcode 0x4129, 1x Unknown NIC, 1x Ethernet Connection, 1x Ethernet Controller I226-V, 1x Raptor Lake PCH CNVi WiFi, 1x 238.5G SAMSUNG MZVL8256HFJD-00BLL, Ubuntu 24.04.4 LTS, 6.17.0-14-generic. | As measured by Intel® Automated Self-Checkout Reference Implementation v3.6.4 | As of February 2026 |
| The Power per stream was 5.4 watts of power per stream on Intel® Core™ 7 150U (RPL-U) CPU and 1.3 watts/stream on iGPU with batch size 1 with the Vision AI pipeline | Processor: Intel Core 7 150U TDP=15W; tested in LENOVO ThinkEdge SE30n; Memory: 32GB (1x32GB DDR5 5600MT/s [5200MT/s]), BIOS M6OKT02A, microcode 0x4129, 1x Unknown NIC, 1x Ethernet Connection, 1x Ethernet Controller I226-V, 1x Raptor Lake PCH CNVi WiFi, 1x 238.5G SAMSUNG MZVL8256HFJD-00BLL, Ubuntu 24.04.4 LTS, 6.17.0-14-generic. | As measured by Intel® Automated Self-Checkout Reference Implementation v3.6.4 | As of February 2026 |