local.aiEarly access
HardwareModelsEnergyWallReferralsContact

Search

Find people, models, hardware, or pages

Account
HardwareModelsEnergyWallReferralsContactClaim your nameAccount
Model

Gpt Oss 20B

Intelligence rank
#86/238
48
Speed rank
#16/238
39s
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.

best41444813.8GB12.9GB12.1GBDisk Size (lower →)Intelligence (higher →)mxfp4-q8 · 48% intelligence · 12.1GB disk sizemxfp4-q8mxfp4 · 42% intelligence · 13.8GB disk sizemxfp4mxfp4 · Dflash · k=7 · 41% intelligence · 13.8GB disk sizemxfp4 · Dflash · k=7
Intelligence Density

Intelligence ÷ Disk Size (GB).

  1. 3.97
    mxfp4-q8
  2. 3.09
    mxfp4
  3. 2.96
    mxfp4 · Dflash · k=7
Hardware

Cost / task speed

Hardware cost / task speed
3 of 3 variations

Model variations

·
4047
Any intelligence
best39s2m 18s8m 01s$16.3k$6.6k$2.7kHardware cost (lower →)Task time (faster →)mxfp4 (42) · NVIDIA DGX Spark 128GB · vllm · $4.7k · 3m 53sSparkmxfp4 (42) · NVIDIA GeForce RTX 4090 24GB · vllm · $3.2k · 58sRTX 4090mxfp4 (42) · NVIDIA GeForce RTX 5090 32GB · vllm · $6.4k · 39sRTX 5090mxfp4 (42) · NVIDIA RTX 6000 Ada 48GB · vllm · $8.2k · 1m 01sRTX6k Adamxfp4 (42) · NVIDIA RTX PRO 6000 Blackwell 96GB · vllm · $16.3k · 41sRTXPro6kmxfp4 (42)mxfp4-q8 (48) · Mac Studio M3 Ultra 96GB · 80-core GPU · vllm · $6.8k · 3m 16sM3 Ultra (80C)mxfp4-q8 (48) · Mac Studio M3 Ultra 96GB · 60-core GPU · vllm · $5.3k · 3m 47sM3 Ultra (60C)mxfp4-q8 (48) · MacBook Pro M4 Max 128GB · 40-core GPU · vllm · $5.6k · 4m 26sM4 Max (40C)mxfp4-q8 (48) · Mac mini M4 Pro 48GB · 20-core GPU · vllm · $2.7k · 8m 01sM4 Pro (20C)mxfp4-q8 (48) · MacBook Pro M5 Max 128GB · 40-core GPU · vllm · $6.7k · 2m 30sM5 Max (40C)mxfp4-q8 (48) · MacBook Pro M5 Pro 64GB · 20-core GPU · vllm · $3.7k · 4m 17sM5 Pro (20C)mxfp4-q8 (48)mxfp4 · Dflash · k=7 (41) · NVIDIA DGX Spark 128GB · vllm · $4.7k · 4m 01sSparkmxfp4 · Dflash · k=7 (41)

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 benchmarksLive benchmark data
ContactAccountReferralsAbout