Model pages · hot models · rapid-mlx 1.1.0

inkling-2-mini-4bit

Run Inkling 2 Mini on a Mac

Run it

Install: pip install rapid-mlx or Rapid-MLX Desktop.

rapid-mlx serve inkling-2-mini-4bit

Minimum RAM: No published minimum; leave room for weights, macOS and context. Run a coding agent on Mac with rapid-mlx launch <agent> (Codex, Claude Code, OpenCode).

Native text-only support in Rapid-MLX 1.1.0. The alias selects the bundled looped implementation and inkling parsers: reasoning is separate from the answer, and streaming and non-streaming tool calls use OpenAI tool_calls. Image input and speculative decoding are unavailable.

Text only · hybrid MoE · inkling2_loopText chat · reasoning · tool calls

Not yet measured on this version
—
on rapid-mlx 1.1.0 (current) · see “Current-version run status”
Weights
—
Peak memory
—

First request

rapid-mlx serve inkling-2-mini-4bit

Weights download on first run; you get an OpenAI-compatible endpoint at http://localhost:8000/v1. No rapid-mlx yet? It's one line — curl -fsSL https://rapidmlx.com/install.sh | bash — or take the desktop app.

Point a client at it. OpenAI SDK: base_url="http://localhost:8000/v1", model="inkling-2-mini-4bit". Claude Code: ANTHROPIC_BASE_URL=http://localhost:8000 (no /v1 suffix) — see the Claude Code guide.

Current-version run status

Benchmark status · No site benchmark

Decode
—
First token
—
Peak memory
—
Weights on disk
—

Source and status

MachineDecodeFirst tokenPeak memoryCold bootWeights
No site benchmark rapid-mlx 1.1.0 · 2026-11-21 · benchmark pending engine alias registry— tok/s————

No site benchmark has been run. Capabilities and the checkpoint revision come from the tagged engine alias registry.

Not yet measured on rapid-mlx 1.1.0. No site benchmark has been run for this build. No decode speed, time to first token or peak memory is claimed. The tagged registry pins mlx-community/Inkling-2-Mini-4bit at 9856dc53edead71405f8348930937c3638ff21a2. Engine alias registry →

Will it fit your Mac?

Peak memory
—
Weights on disk
—

No minimum memory is published in this alias registry; allow room for weights, macOS and context. No site memory measurement is available.

For the conservative install-default placement see the hardware tiers table; to compare against every model your RAM can hold, use the live picker.

Variants & profiles

  • inkling-2-mini-4bit — mlx-community/Inkling-2-Mini-4bit · revision 9856dc53edead71405f8348930937c3638ff21a2

FAQ

How much memory does Inkling 2 Mini need on a Mac?

No minimum memory is published in this alias registry; allow room for weights, macOS and context. No site memory measurement is available.

How fast is Inkling 2 Mini on Apple Silicon?

No site benchmark has been run for this build. No decode speed, time to first token or peak memory is claimed. The tagged registry pins mlx-community/Inkling-2-Mini-4bit at 9856dc53edead71405f8348930937c3638ff21a2.

How do I run Inkling 2 Mini locally?

rapid-mlx serve inkling-2-mini-4bit — the weights download on first run and you get an OpenAI-compatible endpoint at localhost:8000/v1 for Aider, Codex and any OpenAI client, plus the Anthropic API at localhost:8000 for Claude Code.

Where next