local.ailocal.aiEarly access
HardwareModelsEnergyWallReferralsContact
HardwareModelsEnergyWallReferralsContactClaim your nameAccount

Search

Find people, models, hardware, or pages

Account
Model

Gemma 4 E2B It

Intelligence rank
#106/238
21
Speed rank
#3/238
8s
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.

best1820219.2GB6.2GB3.2GBDisk Size (lower →)Intelligence (higher →)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF · 18% intelligence · 3.2GB disk sizeUD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUFUD-Q8_K_XL · 21% intelligence · 5.3GB disk sizeUD-Q8_K_XLfp8 · 18% intelligence · 9.2GB disk sizefp8fp8 · Draft Model · k=4 · 18% intelligence · 9.2GB disk sizefp8 · Draft Model · k=4
Intelligence Density

Intelligence ÷ Disk Size (GB).

  1. 5.59
    UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF
  2. 4.03
    UD-Q8_K_XL
  3. 1.95
    fp8
  4. 1.95
    fp8 · Draft Model · k=4
Hardware

Cost / task speed

Hardware cost / task speed
4 of 4 variations

Model variations

·
1721
Any intelligence
best8s21s56s$15.5k$6.5k$2.7kHardware cost (lower →)Task time (faster →)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · NVIDIA DGX Spark 128GB · llamacpp · $4.7k · 28sSparkUD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · Mac Studio M3 Ultra 96GB · 80-core GPU · llamacpp · $6.8k · 23sM3 Ultra (80C)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · Mac Studio M3 Ultra 96GB · 60-core GPU · llamacpp · $5.3k · 26sM3 Ultra (60C)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · MacBook Pro M4 Max 36GB · 32-core GPU · llamacpp · $3.2k · 28sM4 Max (32C)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · Mac mini M4 Pro 48GB · 20-core GPU · llamacpp · $2.7k · 39sM4 Pro (20C)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · MacBook Pro M5 Max 128GB · 40-core GPU · llamacpp · $6.7k · 19sM5 Max (40C)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · MacBook Pro M5 Pro 64GB · 20-core GPU · llamacpp · $3.7k · 29sM5 Pro (20C)UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · NVIDIA GeForce RTX 4090 24GB · llamacpp · $3.4k · 12sRTX 4090UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · NVIDIA GeForce RTX 5090 32GB · llamacpp · $4.8k · 8sRTX 5090UD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · NVIDIA RTX 6000 Ada 48GB · llamacpp · $8.2k · 12sRTX6k AdaUD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18) · NVIDIA RTX PRO 6000 Blackwell 96GB · llamacpp · $15.5k · 9sRTXPro6kUD-Q4_K_XL · unsloth/gemma-4-E2B-it-GGUF (18)fp8 · Draft Model · k=4 (18) · NVIDIA DGX Spark 128GB · vllm · $4.7k · 24sSparkfp8 · Draft Model · k=4 (18) · NVIDIA GeForce RTX 4090 24GB · vllm · $3.4k · 10sRTX 4090fp8 · Draft Model · k=4 (18) · NVIDIA GeForce RTX 5090 32GB · vllm · $4.8k · 9sRTX 5090fp8 · Draft Model · k=4 (18) · NVIDIA RTX 6000 Ada 48GB · vllm · $8.2k · 10sRTX6k Adafp8 · Draft Model · k=4 (18) · NVIDIA RTX PRO 6000 Blackwell 96GB · vllm · $15.5k · 8sRTXPro6kfp8 · Draft Model · k=4 (18)fp8 (18) · NVIDIA DGX Spark 128GB · vllm · $4.7k · 43sSparkfp8 (18) · NVIDIA GeForce RTX 4090 24GB · vllm · $3.4k · 15sRTX 4090fp8 (18) · NVIDIA GeForce RTX 5090 32GB · vllm · $4.8k · 11sRTX 5090fp8 (18) · NVIDIA RTX 6000 Ada 48GB · vllm · $8.2k · 15sRTX6k Adafp8 (18) · NVIDIA RTX PRO 6000 Blackwell 96GB · vllm · $15.5k · 11sRTXPro6kfp8 (18)UD-Q8_K_XL (21) · NVIDIA DGX Spark 128GB · llamacpp · $4.7k · 43sSparkUD-Q8_K_XL (21) · Mac Studio M3 Ultra 96GB · 80-core GPU · llamacpp · $6.8k · 31sM3 Ultra (80C)UD-Q8_K_XL (21) · Mac Studio M3 Ultra 96GB · 60-core GPU · llamacpp · $5.3k · 31sM3 Ultra (60C)UD-Q8_K_XL (21) · MacBook Pro M4 Max 36GB · 32-core GPU · llamacpp · $3.2k · 40sM4 Max (32C)UD-Q8_K_XL (21) · Mac mini M4 Pro 48GB · 20-core GPU · llamacpp · $2.7k · 56sM4 Pro (20C)UD-Q8_K_XL (21) · MacBook Pro M5 Max 128GB · 40-core GPU · llamacpp · $6.7k · 27sM5 Max (40C)UD-Q8_K_XL (21) · MacBook Pro M5 Pro 64GB · 20-core GPU · llamacpp · $3.7k · 44sM5 Pro (20C)UD-Q8_K_XL (21) · NVIDIA GeForce RTX 4090 24GB · llamacpp · $3.4k · 17sRTX 4090UD-Q8_K_XL (21) · NVIDIA GeForce RTX 5090 32GB · llamacpp · $4.8k · 11sRTX 5090UD-Q8_K_XL (21) · NVIDIA RTX 6000 Ada 48GB · llamacpp · $8.2k · 16sRTX6k AdaUD-Q8_K_XL (21) · NVIDIA RTX PRO 6000 Blackwell 96GB · llamacpp · $15.5k · 12sRTXPro6kUD-Q8_K_XL (21)

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