lean-agent-middle

Middleweight class · v1.0 · Qwen 3.5 base

Advanced agent - complex orchestration, long-context workflows

Verified vs llama.cppPaid

122B total, 10B active per token. A 75 GB model (Q4_K_M) that runs on a single 24 GB GPU through expert offloading - massive knowledge with efficient per-token compute.

Specifications

Total params

122B

Active per token

10B

Base model

Qwen3.5-122B-A10B

Architecture

GDN hybrid MoE

Experts

256 experts, 8 active

Min VRAM / RAM

24 GB / 32 GB

Measured performance

Real numbers from our reference rig, not projections.

Hardware	Prefill	Decode
1x RTX 3090 (24 GB)	6-11 tok/s	5-6 tok/s
2x RTX 3090	4-8 tok/s	6-8 tok/s

Get the model

Download 75.0 GB - single .lmpack, Q4_K_MPrice $19.99 one-time · launch pricing

Purchases are opening soon — the model is packaged and verified, checkout is being finalized.

$ lean pull lean-agent-middle

After purchase you'll receive a license key by email — run lean login once, then pull as usual. Free updates to purchased models.

Quick start (coming soon)

$ curl -sSf https://leanmodels.ai/install.sh | sh
$ lean pull lean-agent-middle
$ lean run lean-agent-middle

Single binary, 15 MB. No Python, no Docker, no cloud dependency.

The installer is not live yet — binaries ship with our first release.