omm

omm verify

Load an installed model into Ollama or LM Studio, send one short deterministic prompt, and confirm it returns real local text.

01 / 06

Overview

Reach for verify when info's compatibility field says failed or unknown, or you just want first-hand proof a model works before relying on it. Unlike a casual test chat, it asks before loading an unloaded model, releases what it loaded when it's done (unless --keep-loaded says otherwise), and never stores the text it generated — only whether generation succeeded.

02 / 06

Options

Every flag this command accepts, and what it defaults to when you leave it out.

  • <name>—Default: required

    An installed model's filename.

  • --engineollama | lmstudioDefault: auto-picks

    Which runtime to test: ollama or lmstudio. Left out, verify picks automatically.

  • --keep-loaded—Default: off

    Keep the model loaded afterward, only if this command is the one that loaded it.

  • --yes / -y—Default: off

    Load the model without asking first. For scripting.

03 / 06

Examples

From a plain search to something you'd put in a script.

Verify an installed model, asking before loading it if needed.

$ omm verify qwen2.5-0.5b-instruct-q4_k_m.gguf

Verify against a specific runtime, without the load confirmation.

$ omm verify qwen2.5-0.5b-instruct-q4_k_m.gguf --engine ollama --yes

Verify and leave the model loaded afterward.

$ omm verify qwen2.5-0.5b-instruct-q4_k_m.gguf --keep-loaded

04 / 06

A real run

Real omm verify qwen2.5-0.5b-instruct-q4_k_m.gguf run, 2026-08-25, this dev machine — a real deterministic prompt actually sent to the real running Ollama and a real answer required back. "already loaded and preserved" is the real detail this run reported, because the model happened to still be loaded from the omm run capture moments earlier; a cold run reports "test load released" instead.

06 / 06

If something goes wrong

Every message below is one this command actually prints. Find yours, read why it happened, then do the last line.

  1. zzzz-fake-model-xyz is not installed via omm. See `omm list`.
    why
    This name doesn't match anything omm has installed.
    what to do
    Run omm list to see the exact filename omm knows about.
    source
    src/omm/cli.py:5157
  2. --engine must be ollama or lmstudio.
    why
    --engine only accepts ollama or lmstudio for verification — no other runtime has a compatibility check implemented yet.
    what to do
    Use ollama or lmstudio, or omit --engine to let verify pick automatically from what the model is linked into.
    source
    src/omm/cli.py:5081-5163

Still stuck? Open an issue with the exact message you saw.

All commands