* test(relayconvert): add golden snapshot matrix and relaykit boundary guard
Phase 0 of the relaykit extraction plan: pin byte-level output of every
registered (from,to) request/response/stream conversion route, and
forbid kit-bound packages from growing host-only imports.
* wip(relayconvert): drop gin.Context from converter signatures; add convmeta draft
Phase 1 in progress: relayconvert now takes context.Context; host media
resolver adapts gin.Context back at the service boundary.
* refactor(relayconvert): decouple converters from RelayInfo, gin, and settings
Phase 1 of the relaykit extraction plan:
- converters now depend on convmeta.Meta (implemented by RelayInfo) instead
of *relaycommon.RelayInfo; ClaudeConvertInfo and the format guesser move
to convmeta with aliases left behind
- host settings reach converters via a convmeta.Options snapshot built in
RelayInfo.ConvOptions; no more model_setting/reasoning global reads inside
the conversion layer
- effort-suffix helpers move to service/relayconvert/reasoning (old package
forwards); chat-to-responses upgrade policy moves to service (host routing
logic, not conversion)
- golden conversion matrix unchanged
* test(relayconvert): tighten boundary — kit packages now free of gin/setting imports
* refactor(dto): drop gin and logger dependencies
Phase 2 (part 1): dto.Request.IsStream now takes *http.Request instead of
*gin.Context (Gemini's impl reads query/path off the std request); dto's
three logger calls become common.SysError. Boundary test allowlist is now
empty — kit-bound packages import no gin/setting/logger/model.
* refactor(kit): extract dependency-free kitutil; dto/types/relayconvert stop importing common
Phase 2 of the relaykit extraction plan:
- new service/relayconvert/kitutil holds the pure helpers the kit needs
(JSON wrappers, pointer/string/uuid/timestamp utils, MaskSensitiveInfo,
pluggable LogInfo/LogError hooks, Debug flag)
- dto, types, and all relayconvert packages now use kitutil; their only
remaining internal deps are dto/types/constant
- common keeps every original symbol (MaskSensitiveInfo delegates to
kitutil) so host code is untouched; main.go routes kit logging into
common.SysLog/SysError and mirrors DebugEnabled
- golden conversion matrix unchanged
* refactor(kit): move EndpointType/FinishReason to types; OpenRouter dialect via Options
Kit packages (dto/types/relayconvert/reasonmap) no longer import constant:
- EndpointType and finish-reason values live in types; constant re-exports
- the OpenRouter special-case in claude->openai request conversion reads
Options.OpenRouterDialect, set by the host from the channel type;
InitChannelMeta invalidates the cached snapshot on channel switch
* refactor: extract relaykit submodule (dto/types/relayconvert/reasonmap)
Phase 3 of the relaykit extraction plan:
- new go module github.com/QuantumNous/new-api/relaykit containing dto
(minus task family), types, relayconvert (with convmeta/kitutil/reasoning),
and reasonmap; host consumes it via require + replace, go.work for dev
- task-family dto (task/suno/midjourney/video) stays in the host dto
package; dual-consumer host files alias it as taskdto
- relaykit builds and tests standalone (GOWORK=off): no host imports,
no gin, no DB, no settings
- golden conversion matrix unchanged
* build(docker): copy relaykit/go.mod before go mod download
The local-replace submodule's go.mod must exist inside the build context
for the main module graph to resolve.
* fix: address relaykit extraction regressions
* fix: address relaykit review regressions
* docs: document Meta nil receiver contract
* fix(relaykit): fail OpenAI→Claude conversion without max_tokens; reject negative default_max_tokens
The Claude Messages API requires max_tokens (omitting it is a 400
"Field required"), but with a nil Options.Claude.DefaultMaxTokens hook
the converters silently emitted a request the upstream is guaranteed to
reject. Both OpenAI Chat and Responses → Claude conversions now return
sharedclaude.ErrMissingMaxTokens when no path (client value, default
hook, thinking-adapter floor) supplied one. Unreachable in the host,
which always configures the hook.
Host side, claude.default_max_tokens now rejects negative values at the
option API before persisting — they would wrap into huge unsigned values
during conversion. Zero stays allowed: the current API treats
max_tokens: 0 as cache pre-warming.
* fix: make Gemini safety settings read path race-free
* refactor: consolidate relay protocol converters
* refactor relayconvert text converters
* feat: refine relay converters and advanced custom routing
* refactor: enhance logging and add thought signature handling for Gemini requests
* refactor: enhance channel cache and pricing endpoint handling for advanced custom models
* feat: preserve billing usage semantics
* feat: add protocol-aware billing usage
* Delete useless files
* chore: update action versions in workflow files
* chore: update Docker action versions in workflow files
* fix: harden billing usage settlement and hot-path route matching
- estimate Gemini completion tokens locally when billable usageMetadata is
prompt-only but output content was received (e.g. client aborts the stream
before the final chunk), and rebuild the attached billing_usage as estimated
so settlement does not bill zero output tokens
- guard NewClaudeMessagesBillingUsage against all-zero ClaudeUsage, matching
the OpenAI/Gemini constructors, so a zero billing_usage cannot override a
non-zero top-level usage during settlement
- cache compiled advanced-custom route model regexes; they run on the request
hot path and were recompiled per request
- move the effectiveBillingUsage remap to PostTextConsumeQuota only, and
document that calculateTextQuotaSummary expects remapped usage
- document the updatePricingLock -> channelSyncLock lock ordering that
InitChannelCache/CacheUpdateChannel rely on, and the aux-struct pitfall in
GeminiChatResponse.UnmarshalJSON
- quota.go: add missing SettleBilling call in PostWssConsumeQuota
- text_quota.go: gate InjectTieredBillingInfo on tieredBillingApplied bool
instead of tieredResult != nil, so fallback billing still logs metadata
- price.go: remove quotaBeforeGroup == 0 from freeModel condition to avoid
bypassing settlement for output-only expressions
- tiered_settle.go: split cc/cc1h subtraction using UsageSemantic to
distinguish OpenAI vs Claude cache creation token formats
- pricing.go: only set BillingMode when a non-empty expression exists
- useModelPricingEditorState.js: only write billing_mode when
finalBillingExpr is non-empty
- Introduced new fields for billing mode and expression in the Pricing model.
- Implemented dynamic pricing breakdown component to display tiered billing details.
- Updated various components to support and render tiered billing information.
- Enhanced pricing calculation logic to accommodate dynamic pricing scenarios.
- Added tests for new billing expression functionalities and UI components.
Keep the model pricing editor wording aligned with the new price-based UI while exposing cache, image, and audio pricing in the marketplace so users can see the full configured pricing model.
This commit refactors the logging mechanism across the application by replacing direct logger calls with a centralized logging approach using the `common` package. Key changes include:
- Replaced instances of `logger.SysLog` and `logger.FatalLog` with `common.SysLog` and `common.FatalLog` for consistent logging practices.
- Updated resource initialization error handling to utilize the new logging structure, enhancing maintainability and readability.
- Minor adjustments to improve code clarity and organization throughout various modules.
This change aims to streamline logging and improve the overall architecture of the codebase.
This commit introduces a major architectural refactoring to improve quota management, centralize logging, and streamline the relay handling logic.
Key changes:
- **Pre-consume Quota:** Implements a new mechanism to check and reserve user quota *before* making the request to the upstream provider. This ensures more accurate quota deduction and prevents users from exceeding their limits due to concurrent requests.
- **Unified Relay Handlers:** Refactors the relay logic to use generic handlers (e.g., `ChatHandler`, `ImageHandler`) instead of provider-specific implementations. This significantly reduces code duplication and simplifies adding new channels.
- **Centralized Logger:** A new dedicated `logger` package is introduced, and all system logging calls are migrated to use it, moving this responsibility out of the `common` package.
- **Code Reorganization:** DTOs are generalized (e.g., `dalle.go` -> `openai_image.go`) and utility code is moved to more appropriate packages (e.g., `common/http.go` -> `service/http.go`) for better code structure.
Backend:
- Model: Add `icon` field to `model.Model` (gorm: varchar(128)); auto-migrated via GORM.
- Pricing API: Extend `model.Pricing` with `icon` and populate from model meta in `GetPricing()`.
Frontend:
- EditModelModal: Add `icon` input (with @lobehub/icons helper link); wire into init/load/submit flows.
- ModelHeader / PricingCardView: Prefer rendering `model.icon`; fallback to `vendor_icon`; final fallback to initials avatar.
- Models table: Add leading “Icon” column, rendering `model.icon` or `vendor` icon via `getLobeHubIcon`.
Notes:
- Backward-compatible. Existing data without `icon` remain unaffected.
- No manual SQL needed; column is added by AutoMigrate.
Affected files:
- model/model_meta.go
- model/pricing.go
- web/src/components/table/models/modals/EditModelModal.jsx
- web/src/components/table/model-pricing/modal/components/ModelHeader.jsx
- web/src/components/table/model-pricing/view/card/PricingCardView.jsx
- web/src/components/table/models/ModelsColumnDefs.js
Backend (Go)
- Include custom endpoints in each model’s SupportedEndpointTypes by parsing Model.Endpoints (JSON) and appending keys alongside native endpoint types.
- Build a global supportedEndpointMap map[string]EndpointInfo{path, method} by:
- Seeding with native defaults.
- Overriding/adding from models.endpoints (accepts string path → default POST, or {path, method}).
- Expose supported_endpoint at the top level of /api/pricing (vendors-like), removing per-model duplication.
- Fix default path for EndpointTypeOpenAIResponse to /v1/responses.
- Keep concurrency/caching for pricing retrieval intact.
Frontend (React)
- Fetch supported_endpoint in useModelPricingData and propagate to PricingPage → ModelDetailSideSheet → ModelEndpoints.
- ModelEndpoints
- Resolve path+method via endpointMap; replace {model} with actual model name.
- Fix mobile visibility; always show path and HTTP method.
- JSONEditor
- Wrap with Form.Slot to inherit form layout; simplify visual styles.
- Use Tabs for “Visual” / “Manual” modes.
- Unify editors: key-value editor now supports nested JSON:
- “+” to convert a primitive into an object and add nested fields.
- Add “Convert to value” for two‑way toggle back from object.
- Stable key rename without reordering rows; new rows append at bottom.
- Use Row/Col grid for clean alignment; region editor uses Form.Slot + grid.
- Editing flows
- EditModelModal / EditPrefillGroupModal use JSONEditor (editorType='object') for endpoint mappings.
- PrefillGroupManagement renders endpoint group items by JSON keys.
Data expectations / compatibility
- models.endpoints should be a JSON object mapping endpoint type → string path or {path, method}. Strings default to POST.
- No schema changes; existing TEXT field continues to store JSON.
QA
- /api/pricing now returns custom endpoint types and global supported_endpoint.
- UI shows both native and custom endpoints; paths/methods render on mobile; nested editing works and preserves order.
Summary
-------
1. Pricing generation
• `model/pricing.go`: skip any model whose `status != 1` when building
`pricingMap`, ensuring disabled models are never returned to the
front-end.
2. Cache refresh placement
• `controller/model_meta.go`
– Removed `model.RefreshPricing()` from pure read handlers
(`GetAllModelsMeta`, `SearchModelsMeta`).
– Kept refresh only in mutating handlers
(`Create`, `Update`, `Delete`), guaranteeing data is updated
immediately after an admin change while avoiding redundant work
on every read.
Result
------
Front-end no longer receives information about disabled models, and
pricing cache refreshes occur exactly when model data is modified,
improving efficiency and consistency.
Summary
-------
This commit unifies soft-delete behaviour across meta tables and
introduces an in-memory cache for model pricing look-ups to improve
throughput under high concurrency.
Details
-------
Soft-delete consistency
• PrefillGroup / Vendor / Model
– Added `gorm.DeletedAt` field with `json:"-" gorm:"index"`.
– Replaced plain `uniqueIndex` with partial unique indexes
`uniqueIndex:<name>,where:deleted_at IS NULL`
allowing duplicate keys after logical deletion while preserving
uniqueness for active rows.
• Imports updated to include `gorm.io/gorm`.
• JSON output now hides `deleted_at`, matching existing tables.
High-throughput pricing cache
• model/pricing.go
– Added thread-safe maps `modelEnableGroups` & `modelQuotaTypeMap`
plus RW-mutex for O(1) access.
– `updatePricing()` now refreshes these maps alongside `pricingMap`.
• model/model_extra.go
– Rewrote `GetModelEnableGroups` & `GetModelQuotaType` to read from
the new maps, falling back to automatic refresh via `GetPricing()`.
Misc
• Retained `RefreshPricing()` helper for immediate cache invalidation
after admin actions.
• All modified files pass linter; no breaking DB migrations required
(handled by AutoMigrate).
Result
------
– Soft-delete logic is transparent, safe, and allows record “revival”.
– Pricing-related queries are now constant-time, reducing CPU usage and
latency under load.
- **Backend Changes:**
- Refactor pricing API to return separate vendors array with ID-based model references
- Remove redundant vendor_name/vendor_icon fields from pricing records, use vendor_id only
- Add vendor_description to pricing response for frontend display
- Maintain 1-minute cache protection for pricing endpoint security
- **Frontend Data Flow:**
- Update useModelPricingData hook to build vendorsMap from API response
- Enhance model records with vendor info during data processing
- Pass vendorsMap through component hierarchy for consistent vendor data access
- **UI Component Replacements:**
- Replace PricingCategories with PricingVendors component for vendor-based filtering
- Replace PricingCategoryIntro with PricingVendorIntro in header section
- Remove all model category related components and logic
- **Header Improvements:**
- Implement vendor intro with real backend data (name, icon, description)
- Add text collapsible feature (2-line limit with expand/collapse functionality)
- Support carousel animation for "All Vendors" view with vendor icon rotation
- **Model Detail Modal Enhancements:**
- Update ModelHeader to use real vendor icons via getLobeHubIcon()
- Move tags from header to ModelBasicInfo content area to avoid SideSheet title width constraints
- Display only custom tags from backend with stringToColor() for consistent styling
- Use Space component with wrap property for proper tag layout
- **Table View Optimizations:**
- Integrate RenderUtils for description and tags columns
- Implement renderLimitedItems for tags (max 3 visible, +x popover for overflow)
- Use renderDescription for text truncation with tooltip support
- **Filter Logic Updates:**
- Vendor filter shows disabled options instead of hiding when no models match
- Include "Unknown Vendor" category for models without vendor information
- Remove all hardcoded vendor descriptions, use real backend data
- **Code Quality:**
- Fix import paths after component relocation
- Remove unused model category utilities and hardcoded mappings
- Ensure consistent vendor data usage across all pricing views
- Maintain backward compatibility with existing pricing calculation logic
This refactor provides a more scalable vendor-based architecture while eliminating
data redundancy and improving user experience with real-time backend data integration.
Summary
• Migrated all ratio-related sources into `setting/ratio_setting/`
– `model_ratio.go` (renamed from model-ratio.go)
– `cache_ratio.go`
– `group_ratio.go`
• Changed package name to `ratio_setting` and relocated initialization (`ratio_setting.InitRatioSettings()` in main).
• Updated every import & call site:
– Model / cache / completion / image ratio helpers
– Group ratio helpers (`GetGroupRatio*`, `ContainsGroupRatio`, `CheckGroupRatio`, etc.)
– JSON-serialization & update helpers (`*Ratio2JSONString`, `Update*RatioByJSONString`)
• Adjusted controllers, middleware, relay helpers, services and models to reference the new package.
• Removed obsolete `setting` / `operation_setting` imports; added missing `ratio_setting` imports.
• Adopted idiomatic map iteration (`for key := range m`) where value is unused.
• Ran static checks to ensure clean build.
This commit centralises all ratio configuration (model, cache and group) in one cohesive module, simplifying future maintenance and improving code clarity.
- Introduce `SelfUseModeEnabled` setting to allow flexible model ratio configuration
- Update error handling to provide more informative messages when model ratios are not set
- Modify pricing and relay logic to support self-use mode
- Add UI toggle for enabling self-use mode in operation settings
- Implement fallback mechanism for model ratios when self-use mode is enabled
CaIon