fix(catalog): repoint three model rows at URLs that resolve

[Siddhesh2377]

Rebuilds the catalog against the current model generation, following the rules agreed in Slack: drop superseded families, keep only 1-bit, Q4 and Q8, and add the newest families across each modality.

Removed: Qwen2, Qwen2.5, Qwen3, LFM2, Llama 2, Llama 3.2, Mistral 7B and SmolLM2 rows, plus the Q2 and Q6 rows that fell outside the allowed quantizations.

Added from a verified list: Qwen3.5 (0.8B through 9B), Gemma 4, Granite 4.1, LFM2.5, the PrismML Bonsai 1-bit family, and Maple Preview.

Every download size is the real Content-Length of the asset rather than a rounded guess, and every URL was checked with a HEAD request. Several sizes in the old catalog were wrong by enough to break the download progress bar.

Private rows are untouched. HNPU/QHexRT and NeuRT bundles are exactly as they were.

Maple Preview is listed as 1-bit, not Q4. Its experts are already ternary in storage, so upstream ships only TQ1_0 and TQ2_0 and calls the uniform Q4_K_M a quantizer sanity check.

LoRA registration is removed for now. The only adapter we ship is trained for qwen2.5-0.5b, which this catalog no longer carries, so registering it would offer an adapter with no compatible base. A replacement adapter trained on a base we do ship is planned, and re-adding is a small edit.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: dd0de598-e7b4-4f82-9b82-818ffec98666

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}