From fa4ba8d9897dda2de8fb4ce081e0705479595b49 Mon Sep 17 00:00:00 2001
From: inference-bot <bot@inference.coop>
Date: Fri, 6 Nov 2026 20:35:04 +0000
Subject: [PATCH] =?UTF-8?q?models:=20document=20greenpt/qwen3-embedding-8b?=
 =?UTF-8?q?=20=E2=80=94=20the=20member-facing=20embeddings=20API?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

New gateway model (code/litellm 54925ca): /v1/embeddings, GreenPT-hosted,
4096-dim (measured through the gateway), $0.25/1M input from GreenPT's own
pricing, 32K input cap (provider catalog). Not a chat model — API-only like
the audio models, hidden from the chat picker. Also states plainly that
Tinfoil's nomic-embed-text is the operator-only docs-RAG embedding, since a
member read its absence from the model list as removal.
---
 models.md | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/models.md b/models.md
index 92fecde..5bdc2de 100644
--- a/models.md
+++ b/models.md
@@ -54,6 +54,9 @@ that **you choose** which axis matters more for a given task.
 - **Cheap and cheerful** → `tinfoil/gpt-oss-120b` or `greenpt/glm-5.3-flash`.
 - **Model origins matter to you** → the Apertus models (publicly developed;
   see below).
+- **Semantic search / RAG (API)** → `greenpt/qwen3-embedding-8b` — the
+  embedding model, on renewable energy (API-only; see [Embedding
+  models](#embedding-models) below).
 
 ## Current models
 
@@ -146,6 +149,33 @@ curl https://gateway.inference.coop/v1/chat/completions \
       ]}]}'
 ```
 
+### Embedding models
+
+Embeddings turn text into a numeric vector — the input to semantic search,
+clustering and RAG. They're called on `/v1/embeddings` with your co-op API key
+and are **not chat models**: they never appear in the chat picker, the API is
+the interface (the same shape as the audio models above).
+
+| Model | Provider | Privacy | Green | Dimensions | Max input | Pricing |
+|---|---|---|---|---|---|---|
+| `🌱 greenpt/qwen3-embedding-8b` | GreenPT | policy | ✅ 100% renewable | 4096 (measured) | 32K tokens (per GreenPT's catalog) | $0.25 / 1M input tokens (no output charge) |
+
+```bash
+curl https://gateway.inference.coop/v1/embeddings \
+  -H "Authorization: Bearer ***" \
+  -H "Content-Type: application/json" \
+  -d '{"model": "greenpt/qwen3-embedding-8b", "input": "the text to embed"}'
+```
+
+The response is standard OpenAI shape: `data[0].embedding` holds the vector,
+and `usage.prompt_tokens` is what meters against your allowance at the rate
+above. Embeddings bill input tokens only — there is no output charge.
+
+For completeness: Tinfoil also serves a private embedding model
+(`nomic-embed-text`), but it is **operator-only** — the co-op runs its own docs
+search on it inside the TEE, and member keys cannot call it.
+`greenpt/qwen3-embedding-8b` above is the member-facing embedding model.
+
 ### Image generation
 
 `🌱 regolo/qwen-image` (Regolo, EU-hosted, zero data retention) generates