# OpenAI text and reasoning inventory

Audited local canonical Scalio source on 2026-09-10. This is a code/config inventory, not a production-traffic verification. Environment overrides, persisted agent-run models and Chatwoot installation settings can change effective models. Text/JSON output from image inputs is included; rendering, embeddings and speech synthesis are excluded.

**Default** means no explicit reasoning-effort parameter. It does not mean reasoning is disabled. Shared wrappers explicitly send **low** when the caller omits effort. OpenAI reasoning settings must be read independently of Gemini fallback thinking settings: for example, avatar-script OpenAI calls use low while their Gemini fallback requests high/medium.

| Use case | Model / configured default | Reasoning effort | Inputs | Outputs | Source |
|---|---|---|---|---|---|
| Copilot chat and creative text | gpt-5.6-luna | Default | Conversation, business/thread context, attachments and tools | Replies, scripts, captions, hooks, briefs and tool calls | [Code](/Users/t/Work/scalio-web/src/app/api/app/copilot/chat/route.ts:912) |
| Copilot turn routing and specialist planning | gpt-5.6-luna | Default | Latest request, history, draft and business state | Intent, content-artifact plan, creative funnel, memory or approval decisions | [Code](/Users/t/Work/scalio-web/src/lib/copilot/turn-planner.ts:645) |
| Copilot creative-task resolver | gpt-5.6-luna / gpt-5.5 | Default | Current task, request, conversation and draft/brief state | Canonical creative task; strong tier for forceStrong, pending input, creative_quote or edit_draft | [Code](/Users/t/Work/scalio-web/src/lib/copilot/creative-task.ts:933) |
| Copilot voice transcript normalization | gpt-5.6-luna | Default | Voice transcript and contextual information | Normalized transcript/intent data | [Code](/Users/t/Work/scalio-web/src/lib/copilot/voice/transcript-normalizer.ts:230) |
| Copilot studio-prompt repair | gpt-5.6-luna | Default | Image/video prompt and business brief | Validation, explanation and repaired prompt | [Code](/Users/t/Work/scalio-web/src/lib/copilot/validators/studio-prompt-repair.ts:35) |
| Calendar Stories brief adaptation | gpt-5.6-luna | Default | Calendar idea or custom text, business context, photo summaries and story date | Structured story brief and recommended photo IDs | [Code](/Users/t/Work/scalio-web/src/lib/calendar/stories.service.ts:657) |
| Calendar, video ideas and ads fallback | gpt-5.6-terra | Low | Same business/calendar/ideas/ads prompt and schema used by primary Gemini call | Structured strategy, content, ideas or targeting result | [Code](/Users/t/Work/scalio-web/src/lib/calendar/llm.ts:194) |
| Instagram caption-shortening fallback | gpt-5.6-luna | Low | Existing caption and limit | Shortened caption | [Code](/Users/t/Work/scalio-web/src/lib/calendar/publishCaption.ts:69) |
| Native website generation/edit fallback | gpt-5.6-luna | High | Business/site prompt, current document and conversation where applicable | Structured website draft or edit plan | [Code](/Users/t/Work/scalio-web/src/lib/website-builder/ai/model-router.ts:484) |
| Maya final chat fallback | gpt-5.6-terra | None | Conversation, business/reel state and tool definitions/results | Chat response and tool calls after Google and Kie routes fail | [Code](/Users/t/Work/scalio-web/src/app/api/app/video-agent/chat/route.ts:98) |
| Storyboard compilation fallback | gpt-5.6-luna | Default | Brief, analysed asset facts and previous validation feedback | Storyboard with scenes/dialogue/camera/style | [Code](/Users/t/Work/scalio-web/src/lib/video-agent/compile-storyboard.ts:528) |
| Traditional Studio v2 garment analysis and prompt authoring | gpt-5.6-luna | Low (shared wrapper) | Garment image plus analysis or framing/model/background directives | Garment structure or image prompt | [Code](/Users/t/Work/scalio-web/src/lib/traditional-studio/v2/garment-analysis-v2.ts:116) |
| Traditional Studio reference-pose extraction | gpt-5.6-luna | High | Reference image | Structured pose description | [Code](/Users/t/Work/scalio-web/src/app/api/app/studio/traditional-studio/extract-pose-from-image/route.ts:89) |
| Traditional Studio reference-background extraction | gpt-5.6-luna | High | Reference image | Structured background description | [Code](/Users/t/Work/scalio-web/src/app/api/app/studio/traditional-studio/extract-background-from-image/route.ts:83) |
| Product Studio analysis and shot prompts | gpt-5.6-luna | Low (shared wrapper) | Product/reference images, category, selected shot and brand/output requirements | Product category, suggested ideas and image/video prompts | [Code](/Users/t/Work/scalio-web/src/lib/external-api/services/product-studio-llm.ts:385) |
| Product Studio image-processing and custom-input endpoints | gpt-5.6-luna | Low | Product image or custom photoshoot idea | Product/shot analysis or structured concept | [Code](/Users/t/Work/scalio-web/src/app/api/app/studio/product-studio/process-custom-input/route.ts:95) |
| Product Studio model-prompt builder | gpt-5.6-luna | Low (shared wrapper) | Product/model composition inputs and prompt template | Model/product image prompt | [Code](/Users/t/Work/scalio-web/src/lib/product-studio/model-prompt-builder.ts:104) |
| Product Studio v2 product analysis | gpt-5.6-luna | Low | Product photo | Category, description, material, presentation and visible brand text | [Code](/Users/t/Work/scalio-web/src/lib/product-studio-v2/llm.ts:62) |
| Product Studio v2 job planning | gpt-5.6-luna | High | Product analysis/photo, pack slots, notes, occasion and brand direction | Art direction and per-slot plan | [Code](/Users/t/Work/scalio-web/src/lib/product-studio-v2/llm.ts:203) |
| Product Studio v2 main-image compliance check | gpt-5.6-luna | Low | Generated image and fill target | White-background / overlay / fill verdict and note | [Code](/Users/t/Work/scalio-web/src/lib/product-studio-v2/generate.ts:139) |
| Avatar video ideas | gpt-5.6-luna | Low (shared wrapper) | Idea type, enriched business context and optional product image | Idea titles and descriptions | [Code](/Users/t/Work/scalio-web/src/app/api/app/studio/avatar-video/ideas/generate/route.ts:366) |
| Avatar promotional script and voice-script processing | gpt-5.6-luna | Low (shared wrapper) | Brief/product image or source script/transliteration instruction | Promotional script or native-script text for TTS | [Code](/Users/t/Work/scalio-web/src/lib/inngest/functions/avatarVideoSteps/avatar-ai-fallback.ts:35) |
| Sora template adaptation and prompt enhancement | gpt-5.6-luna | Medium | Template/user prompt and reference images | Adapted/enhanced video prompt | [Code](/Users/t/Work/scalio-web/src/lib/videos/sora-prompt-fallback.ts:11) |
| Standalone video-prompt generation endpoint | gpt-5 | Low (shared wrapper) | User description and optional image | soraPrompt and a short requested analysis-summary field named thinking | [Code](/Users/t/Work/scalio-web/src/app/api/app/vid/generate-prompt/route.ts:71) |
| Voiceover reference-image descriptions | gpt-5.6-luna | Low | Up to five reference images | Short factual descriptions | [Code](/Users/t/Work/scalio-web/src/lib/vid/dub-voiceover.ts:123) |
| Voiceover delivery direction | gpt-5.6-luna | Low | Script and language | Emotion/pace/emphasis/accent direction for TTS | [Code](/Users/t/Work/scalio-web/src/lib/vid/dub-voiceover.ts:157) |
| Voiceover script drafting | gpt-5.6-luna | Medium | Business facts, product/reference images, duration, region and voice gender | Timed promotional voiceover script | [Code](/Users/t/Work/scalio-web/src/lib/vid/dub-voiceover.ts:295) |
| Video upload face-presence check | gpt-5.6-luna | Low | Reference image | containsFace, confidence and reason | [Code](/Users/t/Work/scalio-web/src/app/api/app/vid/guidelines/analyze/route.ts:50) |
| Reel Studio overlay copy | gpt-5.6-luna | High | Tone, notes and optional image | One catchy line, maximum 90 characters | [Code](/Users/t/Work/scalio-web/src/lib/pipelines/prompts/reel-studio/generate-overlay-text.prompt.ts:26) |
| Image-edit request rewriting | gpt-5.6-luna | Low | User's what-went-wrong request and optional image | Precise English edit instruction | [Code](/Users/t/Work/scalio-web/src/lib/gpt5/convert.ts:13) |
| Product enrichment | gpt-5.6-luna | Low (shared wrapper) | Product URL with web-search tool; researched product description | Detailed description, audience segments, targeting/USPs and short description | [Code](/Users/t/Work/scalio-web/src/lib/inngest/functions/product-enricher.ts:123) |
| Title/description regeneration | gpt-5.6-luna | Low (shared wrapper) | Existing content, context and DB-managed prompt | Rewritten content and requested rationale | [Code](/Users/t/Work/scalio-web/src/lib/api/ai-regeneration/regeneration-service.ts:18) |
| Brand-values extraction | gpt-5.6-luna | Default | DB-managed brand-values prompt with store URL | Plain-text brand values | [Code](/Users/t/Work/scalio-web/src/lib/api/product-images/generators/brand-values.ts:57) |
| Internal Creative Pipeline agent | gpt-5.6-terra | Medium | User request, stored conversation, attachments, research/tool results | Structured answer and tools for creative research/actions | [Code](/Users/t/Work/scalio-internal-creativepipeline/src/lib/agent/runtime.ts:113) |
| Internal agent action authorization/scope checks | gpt-5.6-terra | Medium | Request, preceding question, tools and proposed action arguments | Permitted tool/scope decisions | [Code](/Users/t/Work/scalio-internal-creativepipeline/src/lib/agent/action-scope.ts:54) |
| Support Pulse v1 group validation | gpt-5-mini in checked-in worker config; gpt-5-nano code fallback | Low | Grouped support conversations | Validated support issue analysis | [Code](/Users/t/Work/scalio-web/support-pulse/src/index.ts:689) |
| Support Pulse v2 conversation extraction | gpt-5-mini | Low | Support transcript and metadata | Structured issue/priority/customer-risk analysis and summary | [Code](/Users/t/Work/scalio-web/support-pulse/src/v2.ts:351) |
| Support Pulse v2 root-cause aggregation | gpt-5-mini | Low | Issue-bucket summaries, counts and at-risk customer context | Recurring pain, trend, responsible team and follow-up list | [Code](/Users/t/Work/scalio-web/support-pulse/src/v2.ts:616) |
| Chatwoot Scalio support configuration | gpt-5.6-luna | None (Chat Completions compatibility layer) | Support conversation, approved knowledge and tool context | Support replies and tool calls; installation config must be synced to select this model | [Code](/Users/t/Work/scalio-chatwoot/lib/scalio/captain/support_configuration.rb:50) |
| Chatwoot PDF FAQ generation | gpt-4.1-mini | No explicit reasoning setting | PDF/document content | FAQ extraction/validation | [Code](/Users/t/Work/scalio-chatwoot/enterprise/app/services/captain/llm/paginated_faq_generator_service.rb:18) |
| Chatwoot query translation | gpt-4.1-nano | No explicit reasoning setting | Query and target language | Translated query | [Code](/Users/t/Work/scalio-chatwoot/enterprise/app/services/captain/llm/translate_query_service.rb:2) |
| Chatwoot article search-term generation | gpt-4o | No explicit reasoning setting | Article title, description and content | Search terms/snippets as JSON | [Code](/Users/t/Work/scalio-chatwoot/enterprise/app/models/enterprise/concerns/article.rb:63) |

## Model selection and wrapper evidence

- [Copilot defaults](/Users/t/Work/scalio-web/src/lib/copilot/providers.ts:40): fast = gpt-5.6-luna, strong = gpt-5.5. Most AI SDK calls do not configure effort; the creative-task resolver actively selects strong for some states.
- [Structured-text wrapper](/Users/t/Work/scalio-web/src/lib/external-api/services/structured-text.ts:159): default gpt-5.6-luna, reasoning low for GPT-5-family models. Reference images ARE passed as input_image; the action's comment claiming text-only support is stale.
- [generateJson wrapper](/Users/t/Work/scalio-web/src/lib/external-api/services/openai-gpt5.ts:69): default gpt-5, reasoning low. Most callers override the model explicitly.
- [Calendar fallback](/Users/t/Work/scalio-web/src/lib/calendar/llm.ts:40): gpt-5.6-terra, with low effort. Covers brand strategy/visual-system, calendar plans/variants/refresh/copy adaptation, video ideas and ad targeting. Caption shortening overrides the fallback model to Luna.
- [Maya final fallback](/Users/t/Work/scalio-web/src/app/api/app/video-agent/chat/route.ts:98): currently gpt-5.6-terra, not the older gpt-5.6-sol mentioned in comments. [Compatibility setting](/Users/t/Work/scalio-web/src/lib/video-agent/maya-chat-fallback.ts:319) pins reasoningEffort=none.
- [Support Pulse checked-in worker configuration](/Users/t/Work/scalio-web/support-pulse/wrangler.toml:32) selects gpt-5-mini. V1 code fallback is gpt-5-nano; v2 fallback is gpt-5-mini. This does not verify remote runtime values.
- [Chatwoot support sync](/Users/t/Work/scalio-chatwoot/lib/scalio/captain/support_configuration_sync.rb:56) writes Luna into the installation model setting. [Luna compatibility](/Users/t/Work/scalio-chatwoot/lib/llm/open_ai_chat_completions_compatibility.rb:8) sends reasoning_effort=none.

## Older, configurable, script and unreferenced paths

| Model / path | Status |
|---|---|
| gpt-5-mini and gpt-5-mini-2025-08-07 | Also used in offline video-template localization/seeding scripts: [global examples](/Users/t/Work/scalio-web/scripts/vid-template-seeding/generate-global-examples.ts:70), [localization](/Users/t/Work/scalio-web/scripts/vid-template-seeding/localize-inputs.ts:84), [India examples](/Users/t/Work/scalio-web/scripts/vid-template-seeding/generate-india-app-examples.ts:132), [generic examples](/Users/t/Work/scalio-web/scripts/vid-template-seeding/generate-generic-examples.ts:128). They inherit low effort from the shared wrappers unless overridden. |
| gpt-5-chat-latest, gpt-4o-mini, gpt-4-turbo | Accepted model selections in the [generic text action](/Users/t/Work/scalio-web/src/lib/pipelines/actions/openai/generate-text.ts:25); no fixed application workload selecting them was found. |
| gpt-4o-2024-08-06 | Default of the older [OpenAIService structured-text method](/Users/t/Work/scalio-web/src/lib/external-api/services/openai.ts:393); no callers of that method were found in src. The widely used structuredTextService is a different class and defaults to Luna. |
| gpt-4.1 | Generic [Chatwoot model constant](/Users/t/Work/scalio-chatwoot/lib/llm_constants.rb:4). Config/default evidence, not proof that the Scalio support assistant selects it. Web also uses it as an image-generation orchestrator, outside this text-output inventory. |
| gpt-4.1-mini | Chatwoot also uses this as a [generic LLM default](/Users/t/Work/scalio-chatwoot/lib/llm/config.rb:4) for services whose model is installation-configurable. |
| gpt-5.6-luna Product Studio v2 refineSlot | Defined with high effort, but no caller found in src. [Code](/Users/t/Work/scalio-web/src/lib/product-studio-v2/llm.ts:291). Separate from the Gemini-backed refineSlotPrompt flow. |
| gpt-5.6-luna older Traditional Studio analysis/prompting | Legacy files remain: [analysis](/Users/t/Work/scalio-web/src/lib/inngest/functions/traditionalStudioSteps/handleTSResult-analysis.ts:18) uses wrapper low; [prompting](/Users/t/Work/scalio-web/src/lib/inngest/functions/traditionalStudioSteps/handleTSResult-prompting.ts:66) sets medium. The latter is documented in code as having no callers. |
| gpt-5.6-luna older Sora shotlist function | [Function](/Users/t/Work/scalio-web/src/lib/video-chat/generate-sora-shotlist.ts:8) explicitly documents that it has no callers; no explicit effort. |
| GPT realtime/transcription | [OpenAI realtime path](/Users/t/Work/scalio-web/src/lib/copilot/voice/openai-realtime.ts:69) defaults to gpt-realtime-2 and gpt-4o-transcribe. Listed separately because these are voice/transcription rather than ordinary text authoring. |

Scalio Reply uses Gemini through an OpenAI-compatible LiteLLM client. The use of an OpenAI-compatible SDK is not itself an OpenAI-model workload. Likewise, the internal HR assistant uses the OpenAI SDK against Google's endpoint when Gemini is configured.

No inference calls, database queries, deployments or application-code changes were performed for this audit.