Responsible AI · Model register

Gateway models.
Named.

The production gateway routes and capability profiles available to October Health AI and October People. GPT-5.6 Luna powers the low and medium tiers; GPT-5.6 Sol powers high and extrahigh.

Explore the providers Routing verified 18 September 2026 · production gateway discovery

The provider directory

Who powers what.

Production gateway routes and capability profiles available to Panda AI and People Core. Profile availability does not mean every feature actively uses it. Direct realtime voice is outside this inventory. The transparency evaluations cover only the named text, search and moderation paths.

Download the dated routing inventory (JSON)

Registered AI provider

OpenAI

Training opt-out confirmed

Models 12

gpt-5.6-luna
gpt-5.6-sol
text-embedding-3-small
omni-moderation-2024-09-26
tts-1
gpt-4o-mini-transcribe
omni-moderation-latest
gpt-4o-mini-tts
gpt-audio
gpt-audio-mini-2025-12-15
whisper-1
gpt-image-2.5-sunburst-2026-09-08

Where you’ll find them

Primary text routesLuna · Ash · IvyGateway capability profiles

What they help with

GPT-5.6 Luna: low and medium text tiers, including hiring scoring
GPT-5.6 Luna: Luna mental-health and wellbeing support through the companion search route
GPT-5.6 Luna: Ivy diet, nutrition and meal-planning support
GPT-5.6 Sol: high and extrahigh tiers, including the primary evaluation judge
Separate registered profiles for images, audio, transcription, embeddings, vision and moderation

Training use: Opted out contractually
Conversation logging: Disabled
Data retention: Zero retention
Contract: DPA · SCCs

Risk & human oversight

Risk depends on use. Hiring support remains advisory and requires human review; wellbeing and nutrition responses have defined safety boundaries.

Governance policy

Registered AI provider

Google Gemini

Training opt-out confirmed

Models 3

gemini-3.5-flash-lite
gemini-3.8-flash
gemini-3.1-pro-preview

Where you’ll find them

Text route fallbacksRegistered search profile

What they help with

Gemini 3.5 Flash-Lite: low-tier fallback
Gemini 3.8 Flash: medium-tier fallback and a separate registered search profile
Gemini 3.1 Pro Preview: high and extrahigh fallback
Fallback during eligible primary-provider failures

Training use: Opted out contractually
Conversation logging: Disabled
Data retention: Zero retention
Contract: DPA · SCCs

Risk & human oversight

Fallback models require their own coverage. Results from an OpenAI-primary run do not establish the performance of a Gemini fallback.

Governance policy

Data-processing statement

No October conversation data is used for provider training.

This is enforced contractually through a DPA with each provider and verified during the quarterly audit cycle. API-tier access uses zero data retention.

Explore safety & bias testing

Questions about a model or your data?

Our privacy team can help with provider controls, processing and the register.

Contact our privacy team