docs: add Provider Data Use & ZDR page#60
Conversation
Add a Data Governance page documenting what each of the 25 LLM providers available through Eden AI does with customer data: model training, data retention, and Zero Data Retention (ZDR) availability. - Three-tier model (zero retention by default / no training with ZDR on request / trains by default) mirroring the common ZDR framing - Per-provider comparison tables using commercial API terms (not the weaker consumer-app terms) - EU and data-residency guidance, plus China data-sovereignty flags (DeepSeek, ByteDance, Qwen/DashScope, MiniMax) - API vs consumer-product distinction, and how to control exposure on Eden AI (EU endpoint, provider selection, log retention, BYOK) - Compiled from official provider documentation (June 2026) Wired into the Data Governance navigation group in docs.json. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Rework the Provider Data & ZDR page after review: - ZDR is now a definitive Yes/No taken from OpenRouter's published ZDR endpoint list (scoped to the 25 providers Eden AI offers), replacing the unverifiable "on request" wording. 11 of 25 are ZDR-eligible. - Single scannable comparison table (alphabetical) with a per-cell notes list, plus a quick-pick callout of the ZDR providers. - Clearer structure following the docs UX guidance: stat cards, a "show then tell" Mermaid data-flow diagram, section dividers, colored action cards, and tightened copy. - Methodology note explains the ZDR column source and that absence from OpenRouter's list is not the same as unlimited retention. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Remove all references to the third-party ZDR list; the ZDR column now reads as Eden AI's own classification, with a note that "No" should be read alongside Default retention. - Correct the Data location column from our own routing config (edenai-back host_regions.py): Mistral is EU (Paris), OVHcloud is EU (Roubaix); resolve "Not stated" cells (Anthropic, DeepInfra, Cerebras) to the global API endpoints we actually call. - Rewrite the EU residency section to list the providers truly eligible on api.eu.edenai.run (amazon, google, microsoft, mistral, ovhcloud) and clarify that EU-hosted providers like Scaleway and Nebius are not currently routed through the EU endpoint. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Drop the per-cell notes block and the API-vs-consumer section: since Eden AI only ever calls the provider API, the table shows just that result. - Each provider name links to the policy its row is based on, replacing the standalone Sources list. - Trim to four columns (Provider, Trains?, ZDR, Retention) so the table fits without horizontal scroll; location detail stays in the EU residency and Data sovereignty sections. - Remove remaining stored-for reasons (no "abuse" wording left). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Trains? now reads "No" for every provider except DeepSeek and MiniMax, which are marked "May train"; removes the "Depends" value entirely. - Provider names are plain text again (no inline source links). - Google ZDR kept as Yes per request (despite the AI Studio route). - Reading-card and intro reworded; no free/paid-tier wording remains. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A provider that may train on your data cannot be zero-retention, so set MiniMax ZDR to No and update the count (11 -> 10) in the stat card, the ZDR provider list, and the "pick a ZDR provider" card. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Mark Qwen as Zero Data Retention (direct Model Studio API does not store request content) and bump the count back to 11 in the stat card, ZDR list, and "pick a ZDR provider" card. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Cut the decorative and explanatory chrome (stat cards, data-flow diagram, how-to-read cards, explanatory and disclaimer notes, ZDR tip, next steps) and the Data sovereignty section. The page is now: a one-line intro with Eden AI's posture, the provider table, EU residency, and how to reduce exposure. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
WalkthroughAdds a new Data Governance ZDR documentation page with metadata, provider data-handling details, EU residency guidance, and exposure-reduction steps, and adds it to the Data Governance navigation. ChangesData Governance ZDR page
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~3 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@v3/data-governance/zdr.mdx`:
- Around line 54-62: The “EU data residency” and “Reduce your data exposure”
sections in zdr.mdx currently blur two different rules, so update the wording in
the EU endpoint and ZDR bullet list to clearly separate residency constraints
from zero-data-retention constraints. Use the existing “EU endpoint” and “ZDR
providers” bullets as anchors, and make it explicit that any fallbacks must
remain within the same residency set as the chosen endpoint; do not imply that
the ZDR provider list is EU-resident. If needed, split the guidance into
separate bullets or sub-bullets so readers can distinguish EU residency
requirements from ZDR-only provider eligibility.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 5822c153-c9fc-46f3-a656-f31c0dca692c
📒 Files selected for processing (2)
docs.jsonv3/data-governance/zdr.mdx
| ## EU data residency | ||
|
|
||
| To keep data in the EU, use the [EU endpoint](/v3/data-governance/eu-endpoint) (`api.eu.edenai.run`), which routes only through EU-eligible providers. Today those are Amazon Bedrock, Google Vertex, Microsoft Azure, Mistral (Paris), and OVHcloud (Roubaix). Confirm the current list with `api.eu.edenai.run/v3/models`. | ||
|
|
||
| ## Reduce your data exposure | ||
|
|
||
| - **EU endpoint**: restrict processing to EU-eligible providers. | ||
| - **ZDR providers**: Amazon Bedrock, Cerebras, DeepInfra, Fireworks, Google, Groq, Microsoft Azure, Nebius, Perplexity, Qwen, Together. Pin your `model` and `fallbacks` to these. | ||
| - **Logs off by default**: Eden AI stores no content unless you enable log retention. |
There was a problem hiding this comment.
🔒 Security & Privacy | 🟠 Major | ⚡ Quick win
Separate residency guidance from ZDR guidance.
The EU endpoint note and the ZDR list describe different constraints, but the current wording reads like one shared allowlist. Please make it explicit that fallbacks must stay within the same residency set, or split the bullets so readers don't assume any ZDR provider is EU-resident.
Proposed wording
## Reduce your data exposure
- **EU endpoint**: restrict processing to EU-eligible providers.
-- **ZDR providers**: Amazon Bedrock, Cerebras, DeepInfra, Fireworks, Google, Groq, Microsoft Azure, Nebius, Perplexity, Qwen, Together. Pin your `model` and `fallbacks` to these.
+- **ZDR providers**: Amazon Bedrock, Cerebras, DeepInfra, Fireworks, Google, Groq, Microsoft Azure, Nebius, Perplexity, Qwen, Together.
+- **If you need EU residency**: keep `model` and `fallbacks` within the EU-eligible set above.
- **Logs off by default**: Eden AI stores no content unless you enable log retention.📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| ## EU data residency | |
| To keep data in the EU, use the [EU endpoint](/v3/data-governance/eu-endpoint) (`api.eu.edenai.run`), which routes only through EU-eligible providers. Today those are Amazon Bedrock, Google Vertex, Microsoft Azure, Mistral (Paris), and OVHcloud (Roubaix). Confirm the current list with `api.eu.edenai.run/v3/models`. | |
| ## Reduce your data exposure | |
| - **EU endpoint**: restrict processing to EU-eligible providers. | |
| - **ZDR providers**: Amazon Bedrock, Cerebras, DeepInfra, Fireworks, Google, Groq, Microsoft Azure, Nebius, Perplexity, Qwen, Together. Pin your `model` and `fallbacks` to these. | |
| - **Logs off by default**: Eden AI stores no content unless you enable log retention. | |
| ## EU data residency | |
| To keep data in the EU, use the [EU endpoint](/v3/data-governance/eu-endpoint) (`api.eu.edenai.run`), which routes only through EU-eligible providers. Today those are Amazon Bedrock, Google Vertex, Microsoft Azure, Mistral (Paris), and OVHcloud (Roubaix). Confirm the current list with `api.eu.edenai.run/v3/models`. | |
| ## Reduce your data exposure | |
| - **EU endpoint**: restrict processing to EU-eligible providers. | |
| - **ZDR providers**: Amazon Bedrock, Cerebras, DeepInfra, Fireworks, Google, Groq, Microsoft Azure, Nebius, Perplexity, Qwen, Together. | |
| - **If you need EU residency**: keep `model` and `fallbacks` within the EU-eligible set above. | |
| - **Logs off by default**: Eden AI stores no content unless you enable log retention. |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@v3/data-governance/zdr.mdx` around lines 54 - 62, The “EU data residency” and
“Reduce your data exposure” sections in zdr.mdx currently blur two different
rules, so update the wording in the EU endpoint and ZDR bullet list to clearly
separate residency constraints from zero-data-retention constraints. Use the
existing “EU endpoint” and “ZDR providers” bullets as anchors, and make it
explicit that any fallbacks must remain within the same residency set as the
chosen endpoint; do not imply that the ZDR provider list is EU-resident. If
needed, split the guidance into separate bullets or sub-bullets so readers can
distinguish EU residency requirements from ZDR-only provider eligibility.
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Workflows to automatically generate PRs for you. |
Adds a Data Governance page documenting what each provider does with customer data when called through Eden AI's API.
For every LLM provider on Eden AI, the table shows:
It also covers EU data residency (the providers routable through
api.eu.edenai.run: Amazon Bedrock, Google Vertex, Microsoft Azure, Mistral, OVHcloud) and a short "reduce your data exposure" section. Regions are grounded in our routing config (edenai-backhost_regions.py). Wired into the Data Governance nav group indocs.json.🤖 Generated with Claude Code
Summary by CodeRabbit