refactor(utils): Improve formatModelName logic to handle complex model IDs.

[bluelovers]

refactor(utils): Improve formatModelName logic to handle complex model IDs.

import { formatModelName } from "../src/utils/format-model-name";

let ret: string;

[

	'qwen/qwen3-30b-a3b',
	'ollama/gemma4/gemma4-8.0b-q4_k_m.gguf',

	'ollama/gemma4/gemma4-8.0b-q4_k_m.gguf',

	'ollama/gemma4/gemma4-8.0b-q4_k_m_7.0a',
	'ollama/gemma4/gemma4-8.0b-q4_k_m_7.0',

	'qwen3.6-35b-a3b-uncensored-hauhaucs-aggressive',

	'nvidia/nemotron-3-nano-4b',

	'google/gemma-4-e4b',

].forEach(id => {
	ret = formatModelName({ id } as any);
	console.dir(ret);
});

'Qwen3 30B A3B'
'Gemma4 8.0b Q4 K M .gguf'
'Gemma4 8.0b Q4 K M .gguf'
'Gemma4 8.0b Q4 K M 7.0a'
'Gemma4 8.0b Q4 K M 7.0'
'Qwen3.6 35B A3B Uncensored Hauhaucs Aggressive'
'Nvidia Nemotron 3 Nano 4B'
'Google Gemma 4 E4B'

[agustif]

Thank you @bluelovers for the model-display formatting analysis. The 1.0 architecture in #31 intentionally keeps official model IDs unchanged and lets explicit user names override them rather than inferring formatting. This work is credited in CONTRIBUTORS.md. Closing as superseded by the 1.0 design.