local.ailocal.aiEarly access
HardwareModelsEnergyWallReferralsContact
HardwareModelsEnergyWallReferralsContactClaim your nameAccount

Search

Find people, models, hardware, or pages

Account
Model

LFM2.5 8B A1B

Intelligence rank
#109/238
13
Speed rank
#1/238
5s
Fastest hardware: NVIDIA GeForce RTX 5090 32GB.

Quick Start

Variations

Intelligence / Disk Size

Variations

Verified model-weight disk size. Higher intelligence and lower disk size are better.

best9111316.9GB10.9GB4.8GBDisk Size (lower →)Intelligence (higher →)4bit · 9% intelligence · 4.8GB disk size4bitQ8_0 · 13% intelligence · 9.0GB disk sizeQ8_0Default · LiquidAI/LFM2.5-8B-A1B · 10% intelligence · 16.9GB disk sizeDefault · LiquidAI/LFM2.5-8B-A1B
Intelligence Density

Intelligence ÷ Disk Size (GB).

  1. 1.95
    4bit
  2. 1.46
    Q8_0
  3. 0.57
    Default · LiquidAI/LFM2.5-8B-A1B
Hardware

Cost / task speed

Hardware cost / task speed
3 of 3 variations

Model variations

·
913
Any intelligence
best5s13s34s$15.5k$6.5k$2.7kHardware cost (lower →)Task time (faster →)Default · LiquidAI/LFM2.5-8B-A1B (10) · NVIDIA DGX Spark 128GB · vllm · $4.7k · 34sSparkDefault · LiquidAI/LFM2.5-8B-A1B (10) · NVIDIA GeForce RTX 4090 24GB · vllm · $3.4k · 9sRTX 4090Default · LiquidAI/LFM2.5-8B-A1B (10) · NVIDIA GeForce RTX 5090 32GB · vllm · $4.8k · 5sRTX 5090Default · LiquidAI/LFM2.5-8B-A1B (10) · NVIDIA RTX 6000 Ada 48GB · vllm · $8.2k · 9sRTX6k AdaDefault · LiquidAI/LFM2.5-8B-A1B (10) · NVIDIA RTX PRO 6000 Blackwell 96GB · vllm · $15.5k · 5sRTXPro6kDefault · LiquidAI/LFM2.5-8B-A1B (10)Q8_0 (13) · NVIDIA DGX Spark 128GB · llamacpp · $4.7k · 22sSparkQ8_0 (13) · Mac Studio M3 Ultra 96GB · 80-core GPU · llamacpp · $6.8k · 15sM3 Ultra (80C)Q8_0 (13) · Mac Studio M3 Ultra 96GB · 60-core GPU · llamacpp · $5.3k · 16sM3 Ultra (60C)Q8_0 (13) · MacBook Pro M4 Max 36GB · 32-core GPU · llamacpp · $3.2k · 20sM4 Max (32C)Q8_0 (13) · Mac mini M4 Pro 48GB · 20-core GPU · llamacpp · $2.7k · 28sM4 Pro (20C)Q8_0 (13) · MacBook Pro M5 Max 128GB · 40-core GPU · llamacpp · $6.7k · 14sM5 Max (40C)Q8_0 (13) · MacBook Pro M5 Pro 64GB · 20-core GPU · llamacpp · $3.7k · 22sM5 Pro (20C)Q8_0 (13) · NVIDIA GeForce RTX 4090 24GB · llamacpp · $3.4k · 8sRTX 4090Q8_0 (13) · NVIDIA GeForce RTX 5090 32GB · llamacpp · $4.8k · 5sRTX 5090Q8_0 (13) · NVIDIA RTX 6000 Ada 48GB · llamacpp · $8.2k · 9sRTX6k AdaQ8_0 (13) · NVIDIA RTX PRO 6000 Blackwell 96GB · llamacpp · $15.5k · 6sRTXPro6kQ8_0 (13)4bit (9) · Mac Studio M3 Ultra 96GB · 80-core GPU · vllm · $6.8k · 18sM3 Ultra (80C)4bit (9) · Mac mini M4 Pro 48GB · 20-core GPU · vllm · $2.7k · 26sM4 Pro (20C)4bit (9) · MacBook Pro M5 Max 128GB · 40-core GPU · vllm · $6.7k · 13sM5 Max (40C)4bit (9) · MacBook Pro M5 Pro 64GB · 20-core GPU · vllm · $3.7k · 20sM5 Pro (20C)4bit (9)

Each coloured line is that variation’s cost / task-speed Pareto frontier. Global is the frontier across every qualifying variation; every point is still one original measured configuration.

Frontier →the ranked leaderboardAll models →every measured modelHardware →every measured device
local.aiIndependent local-AI benchmarks
AboutReferralsAccountContact
Live benchmark data