chore: sync models with dashboard API

[amaan-ai20]

Summary

Automated model-catalog sync. The dashboard API (https://api-dashboard.zerogpu.ai/api/models) is the source of truth; every value below was taken from it. The API returns 18 models, and the CLI now matches that set exactly. Merging publishes v3.9.0 to npm (minor: chat --model gained a model).

Changes

Corrected

Model	Field	Was	Now
zlm-v2-iab-classify-edge-enriched	price in/out	$0.02 / $0.05	$0.025 / $0.15
all-minilm-l6-v2	price in	$0.5	$0.004
bge-small-en-v1.5	price in	$0.5	$0.004
gpt-oss-120b	parameters (notes)	117B	120B
qwen3-30b-a3b-fp8	parameters (notes)	30.5B	30B
glm-5.2	context (notes)	1M / 1,048,576	262K / 262,144

Renamed

• deepseek-v4-flash → deepseek-v4-flash-0731, repriced to $0.16 / $0.38 — ZGPU_PRICING, test CATALOG, CHAT_MODELS, README + DOCUMENTATION rows and routing sentences, -m example, ADDING_COMMANDS.

Added

• llama-guard-4-12b (Text Generation, $0.18 / $0.18, 163,840-token context) — pricing, chat --model, README + DOCUMENTATION rows, §1 bullet. Routed to Responses, per the skill's endpoint rule: the payload carries a sample_responses_body and no sample_chat_completions_body. The route can be widened later if the platform confirms Chat Completions support.
• zlm-v1-signal-extract (Text Classification, $0.02 / $0.05) — pricing only; no command, per the task mapping.

Removed

The API no longer returns these, so the CLI stops calling them:

• t5-small — pricing dropped.
• zlm-v1-multi-iab-classify — pricing dropped.
• zlm-v1-followup-questions-edge — pricing and docs dropped; the generate_followups command is deleted (src/commands/generateFollowups.ts, its registration in src/cli.ts, its README section and ToC entry, and DOCUMENTATION §4.14 with the following sections renumbered 4.15–4.18 and their anchors updated). Its exemption comment in tests/savings.test.ts is gone too.

ZGPU_FALLBACK is unchanged: glm-5.2 still holds both the highest input and the highest output rate, so the fallback equality test still holds.

Checklist

• npm run lint passes
• npm test passes (56 tests)
• npm run build succeeds
• Documentation updated (README, DOCUMENTATION, ADDING_COMMANDS)
• audit-models.py --strict exits 0 — only NOTE lines (models priced with no command: zlm-v1-moderation-edge, all-minilm-l6-v2, bge-small-en-v1.5, zlm-v1-signal-extract)
• zerogpu chat --help lists exactly the API's 7 Text Generation ids; zerogpu --help no longer offers generate_followups

🤖 Generated with Claude Code

https://claude.ai/code/session_013AKAY2rGGKEJa6bS4VDL4M

---

Generated by Claude Code

[coderabbitai]

Important

• 🔍 Trigger review

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

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Advanced

Run ID: 21920ff1-a4fd-49f9-8706-80951a2393f4

---

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