Neural Lab
Experimental inference testing on consumer hardware. Validating the edge of local intelligence.
Active Lab Fleet
The Veteranonline
GPU: NVIDIA GTX 1080
VRAM: 8GB GDDR5X
CPU: Intel Core i7-8700K
RAM: 64GB DDR4
"Testing older architecture efficiency and quantized small models."
Mobile Commandonline
GPU: NVIDIA RTX 4050 Laptop (6GB)
VRAM: 6GB GDDR6
CPU: AMD Ryzen 5 7535
RAM: 16GB DDR5
"Entry-level laptop inference testing. Hybrid GPU mode."
Compact Strikeronline
GPU: NVIDIA RTX 4060 Laptop (8GB)
VRAM: 8GB GDDR6
CPU: Intel Core i7-13620H
RAM: 32GB DDR5
"Mid-range laptop inference. Hybrid GPU mode leaves VRAM fully available."
Neon Futureonline
GPU: NVIDIA RTX 5070 (12GB)
VRAM: 12GB GDDR7
CPU: Intel Core Ultra 5
RAM: 96GB DDR5
"Next-gen efficiency testing. DLSS 4.0 validation."
Silicon Minionline
GPU: Apple M4 (10-core)
VRAM: Shared
CPU: M4
RAM: 16GB Unified
"Unified memory architecture testing. CoreML optimizations."
Silicon Maxonline
GPU: Apple M4 Max
VRAM: Shared
CPU: M4 Max
RAM: 64GB Unified
"High-performance mobile inference. 64GB UMA limit testing."
Xenon Interceptoronline
GPU: NVIDIA RTX 5060 Laptop
VRAM: 8GB GDDR6
CPU: Intel Core Ultra 9 275HX
RAM: 32GB DDR5
"High-refresh inference. AlienFX enabled."
G15 Workhorseonline
GPU: NVIDIA RTX 4060 Laptop
VRAM: 8GB GDDR6
CPU: Intel Core i7-13650HX
RAM: 16GB DDR5
"Standard gaming laptop baseline. Validation unit."
Predator Oriononline
GPU: NVIDIA RTX 5070 (12GB) + RTX 5060 (8GB)
VRAM: 20GB Combined
CPU: Intel Core Ultra 7 265KF
RAM: 32GB DDR5
"Asymmetric dual-GPU desktop. Evaluates split-tensor performance."
Latest In-House Benchmarks
| Date | Rig | Model | Quantization | Performance |
|---|---|---|---|---|
| 2026-08-28 | Predator Orion | Llama-3.1-8B(CPU Only (Ultra 7)) | Default (Q4) | 11.0 t/s |
| 2026-08-28 | Xenon Interceptor | Llama-3.1-8B(CPU Only (Ultra 9)) | Default (Q4) | 12.7 t/s |
| 2026-08-28 | Neon Future | Llama-3.1-8B(CPU Only (Ultra 5)) | Default (Q4) | 6.9 t/s |
| 2026-08-01 | Predator Orion | Llama-3.1-8B(5070) | Default (Q4) | 108.5 t/s |
| 2026-08-01 | Predator Orion | Llama-3.1-8B(5070 + 5060) | Default (Q4) | 60.1 t/s |
| 2026-08-01 | Predator Orion | DeepSeek-R1-32B(5070 + 5060 + CPU) | Default (Q4) | 5.5 t/s |
| 2026-02-15 | Neon Future | Gemma-3-4.3B(RTX 5070 (12GB)) | Q4_K_M | 141.8 t/s |
| 2026-02-15 | Neon Future | Gemma-3-4.3B (32k Context)(RTX 5070 (12GB)) | Q4_K_M | 122.7 t/s |
| 2026-02-15 | Neon Future | Gemma-3-4.3B (64k Context)(RTX 5070 (12GB)) | Q4_K_M | 109.5 t/s |
| 2026-02-15 | Neon Future | Llama-3.1-8B(RTX 5070 (12GB)) | Default (Q4) | 99.9 t/s |
SPONSORED LAB// AD_SLOT: 1234567890 // FORMAT: AUTO