Add the four models the catalog was missing, including the embeddings tier

[amaan-ai20]

The dashboard model-pricing API (api-dashboard.zerogpu.ai/api/model-pricing/with-model) serves 23 models. The catalog documents 14. This adds the four of the missing nine that are live, callable, and worth a reader's time.

Every request and response example in this PR is a real call against api.zerogpu.ai, not a hand-written sketch.

Text Embedding is a new tier

Nothing in the docs knew embeddings existed. Both models are routable only on /v1/embeddings:

POST /v1/responses {"model":"all-minilm-l6-v2",...} -> 400 model_not_found
POST /v1/chat/completions {"model":"all-minilm-l6-v2",...} -> 400 model_not_found
POST /v1/embeddings {"model":"all-minilm-l6-v2",...} -> 200, 384-dim vector

That is the same shape as the moderation rollout, so it gets the same treatment: an endpoint page (api-reference/embeddings.mdx), a /embeddings path and schemas in zerogpu.openapi.json, a guide page (docs/embeddings.mdx), and a playground per model.

Model	Params	Dims	Max tokens	Price
all-minilm-l6-v2	22.7M	384	256	$0.50 / 1M input, output not billed
bge-small-en-v1.5	33.4M	384	512	$0.50 / 1M input, output not billed

Both return 384-dimensional vectors, so they are interchangeable in an existing index. Batch input ("input": ["a","b"]) works and is documented.

Two text models

Model	Endpoints	Price	Notes
t5-small	/responses, /chat/completions	$0.05 / $0.40	60M params, 512-token window, cheap summarizer for chunked input
zlm-v1-multi-iab-classify	/responses, /chat/completions	$0.05 / $0.40	90M params, returns the enriched IAB envelope for non-English text

Both are added to the model enums in zerogpu.openapi.json and to their task guides (docs/summarization.mdx, docs/text-classification.mdx). llama-3.1-8b-instruct-fast stays the recommendation when a whole document has to land in one pass; zlm-v2-iab-classify-edge-enriched stays the recommendation for English.

Three things worth a maintainer's eye

1. bge-small-en-v1.5 has no favicon.models-favicon.zerogpu.ai/bge-small-en-v1.5/* 404s and the API returns favicon_url: null, so its catalog row and card render without an icon. Everything else in the table has one.
2. The pricing API carries no metadata for either embedding model (description, parameters, max_tokens are all null; bge was created 2026-08-20). Params, dimensions, and context windows come from the upstream model cards (sentence-transformers, BAAI). Worth backfilling the dashboard so the two sources agree.
3. zlm-v1-multi-iab-classify returns a lossy user_intent.name. The real response to the Spanish sample reads "insights on les son los conceptos sicos del" (accents dropped, words truncated). The docs show the response verbatim rather than tidying it. Categories, topics, and keywords are all correct; only that one string is rough.

Not added, and why

Model	Reason
zlm-v1-iab-classify-cloud	Same-family duplicate of zlm-v1-iab-classify-edge
zlm-v1-iab-classify-edge-enriched (v1)	Superseded by the documented v2
zlm-v1-iab-classify-onnx	Returns 500 "Cloud inference URL is not configured for this model"; on-device only
zlm-v1-signal-extract	400 on both documented request shapes, no published description; documenting it would mean inventing its contract

All nine of the undocumented models are display: false in the dashboard, so if that flag is meant to gate the public catalog, say so and I will drop the four back out.

Verification

• All internal links in new and changed pages resolve to a real page or redirect
• Every docs.json nav entry maps to a file on disk
• All OpenAPI and docs.json JSON parses; the zerogpu.openapi.json diff is additive (192 insertions, 2 deletions, both the enum lines)
• No em dashes in new prose, per docs: remove em dashes from moderation model overview #22

Merge order

cliPR pins its price table to this catalog in both directions. The two are independent to merge (the CLI test uses its own transcription), but they should land together so the transcription is not stale on either side.

🤖 Generated with 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: Pro Plus

Run ID: ed700ec4-64ba-41f5-9742-f4a365fdea2b

---

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.}

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
zerogpu	🟢 Ready	View Preview	Aug 21, 2026, 2:04 PM

💡 Tip: Enable Workflows to automatically generate PRs for you.