1 Commits
Author SHA1 Message Date
ed dad815e6c5 docs(search): describe the search and fetch pipeline
Explain the two stages: SearXNG queries DuckDuckGo and Brave, then Keenable fetches the top result pages. Name the SearXNG instance URL and disclose Keenable's logging, retention, processing location, and training use. Correct the claim that web search has no external dependency.
2026-11-19 14:23:27 -05:00
4 changed files with 28 additions and 59 deletions

No files matched your search

+6 -10
View File
@@ -43,7 +43,7 @@ Your **chat history is stored on our server** so you can revisit past conversati
We commit not to read it. But this is a **policy commitment**, not an architectural guarantee. We're transparent about this distinction because it matters. We commit not to read it. But this is a **policy commitment**, not an architectural guarantee. We're transparent about this distinction because it matters.
Note the boundary precisely: enclave models protect your conversation **while it is being processed** — the model provider and we cannot read the exchange at inference time. Stored history is a separate leg: once a conversation is saved, it sits on our server in a form we can technically read. So for sensitive material, the strongest combination available today is a **Temporary Chat with an enclave model** — protected in transit and at inference, and deleted shortly afterwards. Note the boundary precisely: enclave models protect your conversation **while it is being processed** — the model provider and we cannot read the exchange at inference time. Stored history is a separate leg: once a conversation is saved, it sits on our server in a form we can technically read. So for sensitive material, the strongest combination available today is a **Temporary Chat with an enclave model** — protected in transit and at inference, and never stored afterwards.
### If you use the API, nothing you send is stored ### If you use the API, nothing you send is stored
@@ -71,7 +71,7 @@ it is:
The chat has a **Temporary Chat** mode designed exactly for this: The chat has a **Temporary Chat** mode designed exactly for this:
- Click the **Temporary Chat** toggle in the chat header, and the conversation is hidden from your history. It still exists on our server while you use it, so the chat can work; we retain temporary chats as minimally as possible and delete them automatically shortly after creation. It may remain in encrypted backups until those backups age out. Details are in [Infrastructure](../operations/infrastructure.md#chat-retention-and-web-search). - Click the **Temporary Chat** toggle in the chat header, and the conversation is **never saved** — it doesn't appear in your history, it isn't written to our database, and it disappears when you leave the page.
- To make this the default for every new conversation, open **Settings → Conversations** and enable **Default Temporary Chat**. The preference is stored in your browser, so it's yours to change at any time. - To make this the default for every new conversation, open **Settings → Conversations** and enable **Default Temporary Chat**. The preference is stored in your browser, so it's yours to change at any time.
- The trade-off is that there's nothing to revisit: no history, no search, no sync across devices. Use it whenever that trade is right for you — you can mix saved and temporary conversations freely. - The trade-off is that there's nothing to revisit: no history, no search, no sync across devices. Use it whenever that trade is right for you — you can mix saved and temporary conversations freely.
@@ -84,13 +84,9 @@ Our server software keeps standard operational logs (dates, error messages, and
## 5. Search ## 5. Search
Our self-hosted **SearXNG** instance at Our self-hosted **SearXNG** instance at
[search.inference.coop](https://search.inference.coop) handles web search. For [search.inference.coop](https://search.inference.coop) handles web search. It
chat searches it forwards queries to **DuckDuckGo** and **Brave**; those forwards queries to **DuckDuckGo** and **Brave**, which handle them under their
services receive the query from our server and may see it. Search queries and own policies.
results are part of the conversation record, so they follow that conversation's
retention: deleted when you delete a saved chat, or about an hour after a
Temporary Chat is created. Operational details are in
[Infrastructure](../operations/infrastructure.md#chat-retention-and-web-search).
To read the pages behind the results, LibreChat sends the top result URLs to To read the pages behind the results, LibreChat sends the top result URLs to
**Keenable**, a third-party page-fetch service. We use Keenable's keyless public **Keenable**, a third-party page-fetch service. We use Keenable's keyless public
@@ -124,7 +120,7 @@ The models we serve are provided by third parties that change over time, each wi
## 8. Retention and deletion ## 8. Retention and deletion
- Chat history is retained until you delete it. - Chat history is retained until you delete it.
- **Temporary chats are retained as minimally as possible** — hidden from history immediately and deleted automatically shortly after creation. Details are in [Infrastructure](../operations/infrastructure.md#chat-retention-and-web-search). - **Temporary chats are never retained** — they are not written to our database at all (see §4).
- On leaving, your account is deactivated (data preserved for reactivation). - On leaving, your account is deactivated (data preserved for reactivation).
- You may request full deletion at any time: info@inference.coop. - You may request full deletion at any time: info@inference.coop.
+8 -13
View File
@@ -78,23 +78,18 @@ The emoji in the model name marks its **value tier** (the same icons used in the
a zero-data-retention commitment). The 🌱 emoji marks renewable-powered a zero-data-retention commitment). The 🌱 emoji marks renewable-powered
providers, not a specific company. providers, not a specific company.
| Model | Provider | Privacy | Green | Vision | Context | Pricing (EUR per 1M, in / out / cached) | Notes | | Model | Provider | Privacy | Green | Vision | Context | Pricing (in / out / cached) | Notes |
|---|---|---|---|---|---|---|---| |---|---|---|---|---|---|---|---|
| `🌱 greenpt/deepseek-v4.1-flash` | GreenPT | policy | ✅ 100% renewable | ✅ | 1M | €0.36 / €1.45 / €0.04 | **Default.** DeepSeek's multimodal reasoning model, renewable energy; TEE twin below. Serves the **1017 snapshot** (17 Oct 2026) as of 2026-11-16; GreenPT no longer recommends the base release. The id keeps its version-free name so existing configs keep working. | | `🌱 greenpt/deepseek-v4.1-flash` | GreenPT | policy | ✅ 100% renewable | ✅ | 1M | $0.36 / $1.45 / $0.04 | **Default.** DeepSeek's multimodal reasoning model, renewable energy; TEE twin below. Serves the **1017 snapshot** (17 Oct 2026) as of 2026-11-16; GreenPT no longer recommends the base release. The id keeps its version-free name so existing configs keep working. |
| `🔒 tinfoil/deepseek-v4-1-flash` | Tinfoil | architectural | undisclosed | ✅ | 1M | $0.65 / $1.45 / $0.13 | TEE-protected version of the default; image input, tool calling, long docs/agents. | | `🔒 tinfoil/deepseek-v4-1-flash` | Tinfoil | architectural | undisclosed | ✅ | 1M | $0.65 / $1.45 / $0.13 | TEE-protected version of the default; image input, tool calling, long docs/agents. |
| `🔒 tinfoil/glm-5.3` | Tinfoil | architectural | undisclosed | — | 1M | $1.80 / $5.75 / $0.45 | GLM flagship, TEE-protected; strongest private-tier model for agentic coding and long-horizon tasks. | | `🔒 tinfoil/glm-5.3` | Tinfoil | architectural | undisclosed | — | 1M | $1.80 / $5.75 / $0.45 | GLM flagship, TEE-protected; strongest private-tier model for agentic coding and long-horizon tasks. |
| `🔒 tinfoil/gpt-oss-120b` | Tinfoil | architectural | undisclosed | — | 131K | $0.15 / $0.60 / — | Lightweight open-weight reasoning; cheapest private tier. | | `🔒 tinfoil/gpt-oss-120b` | Tinfoil | architectural | undisclosed | — | 131K | $0.15 / $0.60 / — | Lightweight open-weight reasoning; cheapest private tier. |
| `🌱 greenpt/glm-5.3-flash` | GreenPT | policy | ✅ 100% renewable | — | 1M | €0.11 / €0.44 / €0.022 | Fast GLM model, renewable energy. | | `🌱 greenpt/glm-5.3-flash` | GreenPT | policy | ✅ 100% renewable | — | 1M | $0.11 / $0.44 / $0.022 | Fast GLM model, renewable energy. |
| `🌱 greenpt/glm-5.3` | GreenPT | policy | ✅ 100% renewable | — | 1M | €1.10 / €4.40 / €0.275 | GLM flagship — strongest agentic model in the green tier. | | `🌱 greenpt/glm-5.3` | GreenPT | policy | ✅ 100% renewable | — | 1M | $1.10 / $4.40 / $0.275 | GLM flagship — strongest agentic model in the green tier. |
| `🌱 regolo/apertus-70b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | €1.00 / €3.40 / — | Fully open sovereign model (Apache-2.0), large/capable, image understanding. | | `🌱 regolo/apertus-70b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | $1.16 / $3.94 / — | Fully open sovereign model (Apache-2.0), large/capable, image understanding. |
| `🌱 regolo/qwen3.5-122b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | €0.60 / €3.40 / — | Large vision+reasoning MoE (Qwen 3.5, 122B-A10B); 262K context. | | `🌱 regolo/qwen3.5-122b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | $0.70 / $3.94 / — | Large vision+reasoning MoE (Qwen 3.5, 122B-A10B); 262K context. |
| `🌱 regolo/gemma4-31b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | €0.40 / €0.60 / — | Google's dense multimodal Gemma 4 (31B); image input, native function calling. Served from Regolo's EU renewable-powered infrastructure. | | `🌱 regolo/gemma4-31b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | $0.40 / $0.60 / — | Google's dense multimodal Gemma 4 (31B); image input, native function calling. Served from Regolo's EU renewable-powered infrastructure. |
| `🌱 regolo/mistral-small-4-119b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | €0.20 / €0.70 / — | Mistral's hybrid Small 4 (119B) — instruct + reasoning + coding unified; image input, tool calling. Served from Regolo's EU renewable-powered infrastructure. | | `🌱 regolo/mistral-small-4-119b` | Regolo | policy + ZDR | ✅ 100% renewable | ✅ | 262K | $0.20 / $0.70 / — | Mistral's hybrid Small 4 (119B) — instruct + reasoning + coding unified; image input, tool calling. Served from Regolo's EU renewable-powered infrastructure. |
**Currency.** GreenPT and Regolo publish prices in **EUR**; those are the stable
list prices above. The gateway bills in USD, converting at the **ECB daily
reference rate**, so a member's dollar cost follows the exchange rate rather than
a frozen conversion. Tinfoil publishes in USD and is listed as such.
Vision is verified working through the co-op gateway (base64 image input via Vision is verified working through the co-op gateway (base64 image input via
`/v1/chat/completions`, standard OpenAI format). In the chat UI, just attach `/v1/chat/completions`, standard OpenAI format). In the chat UI, just attach
+1 -3
View File
@@ -3,9 +3,7 @@
A central part of our mission is to use collective purchasing power to generate better options for AI service than we have on our own. We have no investors or outside donors. Our co-op depends on its members. A central part of our mission is to use collective purchasing power to generate better options for AI service than we have on our own. We have no investors or outside donors. Our co-op depends on its members.
Usage credits are priced at what the co-op pays providers — see Usage credits are priced at what the co-op pays providers — see
[Models](models.md) for current prices. GreenPT and Regolo list prices are in [Models](models.md) for current prices. What members contribute beyond their usage
EUR; billing converts them to USD at the ECB daily rate, so the dollar figure
follows the exchange rate rather than a frozen conversion. What members contribute beyond their usage
covers the co-op's expenses. covers the co-op's expenses.
## Membership dues ## Membership dues
+13 -33
View File
@@ -75,8 +75,19 @@ code](#the-code) below for where each piece lives.
Open Collective (billing and fiscal sponsorship) is the one external service in Open Collective (billing and fiscal sponsorship) is the one external service in
the membership flow. See the membership flow. See
[opencollective.com/inference-cooperative](https://opencollective.com/inference-cooperative). [opencollective.com/inference-cooperative](https://opencollective.com/inference-cooperative).
Web search uses external services too; see Web search uses external services: DuckDuckGo and Brave receive queries, and
[Chat retention and web search](#chat-retention-and-web-search). Keenable fetches page content. See [Web search](#web-search).
### Web search
LibreChat's web search runs in two stages. A self-hosted **SearXNG** instance at
[search.inference.coop](https://search.inference.coop) sends the query to
DuckDuckGo and Brave. It returns candidate results. LibreChat sends the top
result URLs to **Keenable**, a third-party page-fetch service.
Keenable extracts text from each page. The model uses that text as context for
the answer. SearXNG runs on co-op infrastructure. Keenable is an external
service, and the [privacy policy](../governance/privacy-policy.md#5-search)
describes its logging and retention.
## Mail ## Mail
@@ -174,37 +185,6 @@ The co-op serves models from multiple providers — Tinfoil for architectural
privacy, GreenPT and Regolo for renewable energy. See the [model list](../guides/models.md), privacy, GreenPT and Regolo for renewable energy. See the [model list](../guides/models.md),
the single source of truth for which models are available and how they're named. the single source of truth for which models are available and how they're named.
## Chat retention and web search
**Saved chats** live in the chat database and its search index until the member
deletes them. Web-search tool calls are part of that record: the query and the
results returned are stored with the message.
**Temporary chats** are server-backed — messages must reach the server while the
chat is running — so they are not client-only. The chat is configured to delete
them from the database about **one hour after creation** (`interface.temporaryChatRetention: 1`
in the chat configuration; one hour is the minimum this LibreChat version
accepts). They are hidden from history immediately, and the database enforces the
expiry automatically. Temporary chats created before this setting took effect
keep their original 30-day expiry until they age out.
**Backups are a separate layer.** Cloudron backups are whole-app snapshots and
cannot exclude individual chats, so a temporary chat present when a backup runs
can remain recoverable from that backup until the backup itself expires (see
[Backups](#backups)).
**Web search** in the chat uses **DuckDuckGo and Brave** through the co-op's
self-hosted SearXNG instance. Queries are sent from the co-op server, so the
upstream engines see the query but not the member's IP or identity. The SearXNG
instance is also publicly reachable at search.inference.coop, and its logs can
contain query text when an upstream engine request fails; those log lines carry
no member identity or IP, and are retained only briefly. The reverse proxy does
not log search queries.
LibreChat reads the pages behind the results through **Keenable**, a third-party
page-fetch service. Keenable's logging and retention are described in the
[privacy policy](../governance/privacy-policy.md#5-search).
## Security posture (plain language) ## Security posture (plain language)
- **Member isolation.** Each API key is scoped to its owner's team and budget; a - **Member isolation.** Each API key is scoped to its owner's team and budget; a