Skip to content

Olares One AI benchmarks

Explore LLM, image and video benchmark results for Olares One, with hardware specifications and the software settings used in each test.

Hardware specifications

CPU
Intel Core Ultra 9 275HX
GPU / SoC
NVIDIA GeForce RTX 5090 Laptop
GPU memory (VRAM)
24 GB GDDR7
System memory
96 GB DDR5
INT8 compute
456 TOPS
Memory bandwidth
896 GB/s
Storage
2 TB SSD
OS
Olares OS
Driver
NVIDIA 620.70.06, CUDA 13.5

Benchmark results

ModelBenchmarkConfigurationOlares One
Qwen3.8-27BIFBench
Evaluation time
llama.cpp, NVFP4 LOW8.7 h
Qwen3.8-Flash-NextIFBench
Evaluation time
Strata, IQ2_XS7.7 h
Qwen3.8-27BGPQA Diamond
Evaluation time
llama.cpp, NVFP4 LOW11.5 h
Qwen3.8-27BInference speed
Token generation, single request
llama.cpp, NVFP4 LOW93 tok/s
Qwen3.8-Flash-NextInference speed
Token generation, single request
Strata, Q2_0166 tok/s
Qwen-Image-2.1 BaseText to image
Per image
ComfyUI, Nunchaku FP412 s
Qwen-Image-2.1 BaseImage edit
Per image
ComfyUI, Nunchaku FP417 s
Qwen-Image-2.1 TurboImage edit
Per image
ComfyUI, Nunchaku FP45.2 s
MiniMax H3 FastH3 v2Text to video
Sampling time

720p, 5 s
ComfyUI, nvfp478 s

Test configurations

Qwen3.8-27B

IFBench

llama.cpp, NVFP4 LOW

Engine and build
llama.cpp b11718 (547208d), official prebuilt llama-b11718-bin-ubuntu-cuda-13.4-x64
Launch arguments
--mmproj mmproj-F16.gguf -c 104448 -ngl all -fa on -ctk q8_0 -ctv q8_0 --jinja -np 1 --spec-type draft-mtp --spec-draft-n-max 4

IFBench

llama.cpp, UD-Q4_K_XL

Engine and build
llama.cpp b12394 (07037ba13), official prebuilt llama-b12394-bin-ubuntu-cuda-13.5-x64, with the libs of cudart-llama-b12394-bin-ubuntu-cuda-13.5-x64 copied next to llama-server
Launch arguments
--mmproj mmproj-F16.gguf -c 104448 -ngl all -fa on -ctk q8_0 -ctv q8_0 --jinja -np 1 --spec-type draft-mtp --spec-draft-n-max 5

GPQA Diamond

llama.cpp, NVFP4 LOW

Engine and build
llama.cpp b11718 (547208d), official prebuilt llama-b11718-bin-ubuntu-cuda-13.4-x64
Launch arguments
-c 104448 -ngl all -fa on -ctk q8_0 -ctv q8_0 --jinja -np 2 -kvu --mmproj mmproj-F16.gguf --spec-type draft-mtp --spec-draft-n-max 4

Inference speed

llama.cpp, NVFP4 LOW

Engine and build
llama.cpp b11718 (547208d), official prebuilt llama-b11718-bin-ubuntu-cuda-13.4-x64
Launch arguments
--mmproj mmproj-F16.gguf -c 104448 -ngl all -fa on -ctk q8_0 -ctv q8_0 --jinja -np 4 -kvu --spec-type draft-mtp --spec-draft-n-max 4

Inference speed

llama.cpp, UD-Q4_K_XL

Engine and build
llama.cpp b12394 (07037ba13), official prebuilt llama-b12394-bin-ubuntu-cuda-13.5-x64, with the libs of cudart-llama-b12394-bin-ubuntu-cuda-13.5-x64 copied next to llama-server
Launch arguments
--mmproj mmproj-F16.gguf -c 104448 -ngl all -fa on -ctk q8_0 -ctv q8_0 --jinja -np 4 -kvu --spec-type draft-mtp --spec-draft-n-max 2

Qwen3.8-Flash-Next

IFBench, Inference speed

Strata, IQ2_XS

Engine and build
Strata strata 0.2.10 (engine 0.1.56)
Launch arguments
--pack iq2_xs --native Qwen3.8-Flash-Next-GSQ-RCO-IQ2_XS-00001-of-00002.gguf --ple-gguf Qwen3.8-Flash-Next-GSQ-RCO-IQ2_XS-00002-of-00002.gguf --expert-profile expert-profile.bin --expert-cache auto --prefill auto --spec 4 --spec-min-p 0.5 --mtp rt --max-context 131072 --kv int8 --ple-io ram --pool-workers 15 --vram-reserve-mib 1024 --adapt-every 100000

IFBench, Inference speed

Strata, IQ3_XXS

Engine and build
Strata strata 0.2.10 (engine 0.1.56)
Launch arguments
--pack iq3_xxs --native Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00001-of-00002.gguf --ple-gguf Qwen3.8-Flash-Next-GSQ-RCO-IQ3_XXS-00002-of-00002.gguf --expert-profile expert-profile.bin --expert-cache auto --prefill auto --spec 4 --spec-min-p 0.5 --mtp rt --max-context 131072 --kv int8 --pool-workers 15 --vram-reserve-mib 1024 --adapt-every 100000

IFBench, Inference speed

Strata, UD-IQ4_XS

Engine and build
Strata strata 0.2.10 (engine 0.1.56)
Launch arguments
--pack unsloth-ud-iq4_xs --native Qwen3.8-Flash-Next-UD-IQ4_XS-00001-of-00003.gguf --expert-profile expert-profile.bin --expert-cache auto --prefill auto --spec 4 --spec-min-p 0.5 --mtp rt --max-context 131072 --kv int8 --ple-io ram --resident-budget-gib 56 --pool-workers 15 --vram-reserve-mib 1024 --adapt-every 100000

Inference speed

Strata, Q2_0

Engine and build
Strata strata 0.2.10 (engine 0.1.56)
Launch arguments
--pack q2_0 --native Qwen3.8-Flash-Next-GSQ-RCO-Q2_0-00001-of-00002.gguf --ple-gguf Qwen3.8-Flash-Next-GSQ-RCO-Q2_0-00002-of-00002.gguf --expert-profile expert-profile.bin --expert-cache auto --prefill auto --spec 4 --spec-min-p 0.5 --mtp rt --max-context 131072 --kv int8 --ple-io ram --pool-workers 15 --vram-reserve-mib 1024 --adapt-every 100000

Inference speed

Strata, IQ3_S

Engine and build
Strata strata 0.2.10 (engine 0.1.56)
Launch arguments
--pack iq3_s --native Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00001-of-00002.gguf --ple-gguf Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf --expert-profile expert-profile.bin --expert-cache auto --prefill auto --spec 4 --spec-min-p 0.5 --mtp rt --max-context 131072 --kv int8 --pool-workers 15 --vram-reserve-mib 1024 --adapt-every 100000

Qwen-Image-2.1

ComfyUI, Nunchaku FP4

Qwen-Image-2.1 Base: Text to image, Image edit

Qwen-Image-2.1 Turbo: Image edit

Engine and build
ComfyUI v0.44.0-23-g63e5bf59, git worktree of comfyui at comfyui-nunchaku, venv/comfyui-nunchaku (uv pip, torch 2.14.1 from PyPI), custom node tonera/vitoom-nunchaku-comfy@07ae35d96a621800c5b3b6e6620355b510aea11f, nunchaku wheel 1.4.1+cu13.0torch2.14 from tonera/vitoom-nunchaku@51c61df1, CUDA, torch 2.14.1+cu130, Python 3.12.3
Launch arguments
--listen 127.0.0.1 --port 8288 --extra-model-paths-config extra_model_paths.yaml --output-directory images --input-directory inputs

ComfyUI, NVFP4

Qwen-Image-2.1 Base: Text to image, Image edit

Qwen-Image-2.1 Turbo: Image edit

Engine and build
ComfyUI v0.44.0-23-g63e5bf59, git checkout in comfyui, venv/comfyui (uv pip, PyTorch CUDA wheels), CUDA, torch 2.15.0+cu132, Python 3.12.3
Launch arguments
--listen 127.0.0.1 --port 8288 --extra-model-paths-config extra_model_paths.yaml --output-directory images --input-directory inputs

ComfyUI, int8

Qwen-Image-2.1 Base: Text to image, Image edit

Qwen-Image-2.1 Turbo: Image edit

Engine and build
ComfyUI v0.44.0-23-g63e5bf59, git checkout in comfyui, venv/comfyui (uv pip, PyTorch CUDA wheels), CUDA, torch 2.15.0+cu132, Python 3.12.3
Launch arguments
--listen 127.0.0.1 --port 8288 --extra-model-paths-config extra_model_paths.yaml --output-directory images --input-directory inputs

MiniMax H3

FastH3 v2: Text to video

ComfyUI, nvfp4

Engine and build
ComfyUI v0.44.0-23-g63e5bf59, git checkout in comfyui, venv/comfyui (uv pip, PyTorch CUDA wheels), CUDA, torch 2.15.0+cu132, Python 3.12.3
Launch arguments
--listen 127.0.0.1 --port 8288 --disable-auto-launch --disable-dynamic-vram --disable-pinned-memory --disable-cuda-malloc --use-ck-attention --disable-smart-memory --cache-none --reserve-vram 1 --extra-model-paths-config extra_model_paths.yaml --output-directory output --input-directory input --temp-directory temp --user-directory user, PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True H3_METRICS_DIR=metrics