Skip to main content
Developer platform

FinanceGPT API documentation

Build governed company finance, investment, quantitative, workflow and AI experiences on API v2 with explicit scopes, evidence and execution boundaries.

OpenAPI YAML ML service
Overview Quickstart API reference Guides LQM Builder SDKs Auth & webhooks Changes Versioning CI Support Certification Programme Ecosystem Solutions References Evidence exchange Status Marketplace API plans
Looking for an endpoint? Search the complete API Reference generated from the published OpenAPI contract.
Start here

Getting started

API v2 is workspace-scoped. Sandbox and production credentials are separate.
API v2
Authentication

Bearer API key or OAuth service credential issued from the workspace developer console.

Authorization

Every operation requires an explicit product scope. Missing scopes are denied rather than silently downgraded.

Environment

Sandbox responses are deterministic fixtures and do not read production workspace data.

Authorization: Bearer <FinanceGPT API key or OAuth service token>
Accept: application/json
Governed by default. Data access, model inference and financial execution remain separately permissioned. Documentation does not imply execution authority.

Company Finance

Normalized statements, deterministic ratios, resilience, valuation and scenarios.
Available
MethodEndpointScopePurpose
GET /api/v2/companies financial:analysis:read List company finance workspaces
GET /api/v2/companies/{company}/financials financial:analysis:read Normalized periods and provenance
GET /api/v2/companies/{company}/ratios financial:analysis:read Deterministic ratio library
GET /api/v2/companies/{company}/analysis financial:analysis:read Ratios, resilience, DCF and linked forecast
POST /api/v2/companies/{company}/valuations financial:analysis:run Run explicit-assumption DCF
POST /api/v2/companies/{company}/scenarios financial:analysis:run Run deterministic scenario analysis
Example
curl -H "Authorization: Bearer $FINANCEGPT_API_KEY" \
  -H "Accept: application/json" \
  https://financegpt.uk/api/v2/companies/123/analysis

Data Integrations

Accounting and banking connectors with OAuth, encrypted credentials, synchronization provenance and explicit live verification.
Configuration dependent
Xero

OAuth 2.0, organisation selection, token refresh, financial reports and bank transactions.

QuickBooks Online

OAuth 2.0, report/CDC synchronization, token refresh and signed webhook intake where configured.

Plaid

Link flow, incremental Transactions Sync, balances, item health and webhook-driven refresh where configured.

Verification is explicit. Connector implementation, platform configuration, workspace authorization and the most recent live read check are represented separately. Technical integration does not imply a commercial partnership.

Investment Clients & Portfolios

Manager-scoped client relationships, portfolio records, performance measurement and governed portfolio workflows.
Governed surface

FinanceGPT separates recorded holdings from provider-backed market evidence and from broker execution. Client relationship data remains workspace-scoped and is not exposed through public client APIs unless explicitly governed.

Recorded holdings Provider-backed prices Financial plans TWR performance Portfolio construction Scenario analysis Mandate controls Execution governance Custody reconciliation
MethodEndpointScopePurpose
GET/api/v2/investment/portfolios/{portfolio}/performanceinvestment:readRead TWR runs and evidence-linked valuation subperiods.
POST/api/v2/investment/portfolios/{portfolio}/performance/observationsinvestment:runRecord opening/closing valuation and explicit external cash flow.
POST/api/v2/investment/portfolios/{portfolio}/performance/calculateinvestment:runChain-link non-overlapping subperiod returns and persist methodology evidence.

Financial Data

Canonical provider-backed financial and market data with rights and provenance controls.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Planning

Forecast, budget and scenario operations with explicit assumptions and evidence.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Investment Intelligence

Company, portfolio, risk, research and governance operations.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Investment Execution

Approved order-batch preparation, pre-trade controls and reconciliation.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Quant Models

Deterministic quantitative model catalog and governed model execution.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

LQM Builder

Build, evaluate and emit governed quantitative LQMs with versioned Evidence Packets. Language remains a separate QLM-selected runtime.

An LQM is a quantitative module for forecasting, valuation, risk, simulation, optimization or scenario generation. It does not receive raw document text and does not generate language. QLM Studio composes promoted LQMs with a language runtime, evidence, tools and governance.

MethodEndpointScopePurpose
GET/api/v2/lqm/platformlqm:build:readRead the reconciled LQM platform contract, reused infrastructure, execution boundaries and readiness.
GET/api/v2/lqm/developer-manifestlqm:build:readRead the machine-readable developer contract, least-privilege capability profiles and lifecycle boundaries.
GET/api/v2/lqm/models/{model}/developer-manifestlqm:build:readRead model-specific lifecycle readiness and currently eligible developer operations.
GET/api/v2/lqm/modelslqm:build:readList governed workspace LQM registry records.
POST/api/v2/lqm/modelslqm:build:writeRegister a quantitative LQM identity, task family and intended use before training.
POST/api/v2/lqm/datasets/{dataset}/contractslqm:build:writeSeal a numeric/time-series dataset contract with source, schema and lineage hashes.
POST/api/v2/lqm/models/{model}/training-specslqm:build:writeSeal reproducible VAE-GAN training configuration and lineage before compute begins.
GET/api/v2/lqm/models/{model}/training-runslqm:build:readList governed VAE-GAN training runs and sealed checkpoint metadata.
POST/api/v2/lqm/models/{model}/training-runslqm:build:runQueue a sealed training specification through the private FinanceGPT ML runtime.
GET/api/v2/lqm/training-runs/{run}lqm:build:readRead exact parameter count, checkpoint SHA-256, training metrics and runtime evidence for a run.
GET/api/v2/lqm/models/{model}/evaluationslqm:build:readList statistical-fidelity and robustness evaluations for governed checkpoints.
POST/api/v2/lqm/models/{model}/evaluatelqm:build:runEvaluate a sealed checkpoint against the versioned LQM gate suite before promotion.
GET/api/v2/lqm/evaluation-runs/{evaluation}lqm:build:readRead gate results, sealed evaluation hash, baseline comparison and runtime evidence.
GET/api/v2/lqm/models/{model}/evidence-schemalqm:build:readRead the versioned Evidence Packet JSON contract and its sealed schema SHA-256.
GET/api/v2/lqm/models/{model}/evidence-packetslqm:build:readList immutable typed evidence emitted by evaluated checkpoints.
POST/api/v2/lqm/models/{model}/evidence-packetslqm:build:runRun a bounded evaluated-checkpoint inference and seal its quantitative output as an Evidence Packet.
GET/api/v2/lqm/evidence-packets/{packet}lqm:build:readRead the typed packet, provenance, checkpoint, evaluation and model-ledger hashes.
POST/api/v2/lqm/evidence-packets/{packet}/verifylqm:build:runIndependently re-hash the schema, packet, checkpoint, evaluation, inference output and hash-linked model evidence ledger.
GET/api/v2/lqm/models/{model}/operations-policylqm:build:readRead the model's governed drift/retraining cadence, thresholds, cooldown and review boundary.
PUT/api/v2/lqm/models/{model}/operations-policylqm:build:writeConfigure production training operations without granting automatic checkpoint replacement.
GET/api/v2/lqm/models/{model}/drift-assessmentslqm:build:readList deterministic dataset/model-quality drift evidence for an evaluated LQM checkpoint.
POST/api/v2/lqm/models/{model}/drift-assessmentslqm:build:runCompare a newer governed numeric dataset with the last passing evaluated checkpoint and persist drift evidence.
GET/api/v2/lqm/models/{model}/retraining-requestslqm:build:readList review-gated retraining requests and candidate checkpoint state.
POST/api/v2/lqm/models/{model}/retraining-requestslqm:build:writeCreate a retraining request without starting compute.
POST/api/v2/lqm/retraining-requests/{request}/decisionlqm:build:writeExplicitly approve, reject or cancel a retraining candidate. Approval authorizes candidate training only.
POST/api/v2/lqm/retraining-requests/{request}/advancelqm:build:runAdvance an approved request through training and the required evaluation suite to candidate-ready state; no active-version change occurs.
GET/api/v2/lqm/models/{model}/publicationslqm:build:readRead model-hub publication state, revision identifiers, builder attribution and sealed supply-chain hashes.
POST/api/v2/lqm/models/{model}/publishlqm:publish:writeHuman-confirm publication of an exact evaluated, independently verified checkpoint through the builder's connected publishing identity.
GET/api/v2/lqm/publications/{publication}lqm:build:readRead immutable publication metadata, remote revision, artifact SHA-256 values and supply-chain head.
GET/api/v2/lqm/models/{model}/promotionslqm:build:readRead governed promotion history, exact Quant Models projection versions and decision evidence.
POST/api/v2/lqm/models/{model}/promotelqm:promote:writeHuman-promote an exact passing checkpoint into the workspace Quant Models catalog after Evidence Packet and drift revalidation.
POST/api/v2/lqm/promotions/{promotion}/retirelqm:promote:writeRetire a promoted LQM version. Existing QLM versions fail closed and are never silently rebound.
Training execution rule. Training runs bind an immutable specification to the existing ML training-job control plane and private numeric runtime. The source/configuration lineage is not rewritten after compute starts.
One control plane. LQM Builder extends the existing ML registry, training jobs, quantitative catalog, QLM Studio, API v2 and FinanceGPT AI Credits. It does not create a second router, model registry or credit ledger.
Evidence is the interface. A passing checkpoint may emit financegpt.lqm.evidence.v1 packets. QLM language runtimes may explain those typed outputs; they never receive the LQM latent state or become the source of numeric authority.
Checkpoint is not promotion. Training, passing evaluation and verified Evidence Packet evidence remain non-active until an explicit human promotion decision re-attests the exact checkpoint, evaluation, packet and drift posture.
Retraining is review-gated. Drift and schedule triggers may create retraining requests, but they never silently launch compute or replace active_version_id. A candidate becomes live only through the separate human promotion gate.
Promotion is not publishing. Promotion activates an exact governed projection inside the workspace Quant Models catalog. Model-hub publication uses a separate builder connection, scope, human confirmation and supply-chain record; neither action implies the other.
QLM composition is immutable by version. When a QLM version selects a promoted LQM, FinanceGPT binds the exact promotion hash, checkpoint SHA-256, Evidence Packet schema and Quant Models version. Retirement fails closed; FinanceGPT never silently rebinds an existing QLM version.

QLMs

Governed quantitative-language profiles combining model, evidence and tool boundaries.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Quant Compute

Bounded backtest, simulation and optimization jobs with lineage.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

ML

Approved learned-model inference with separate runtime verification and model governance.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Workflows

Repeatable finance workflows, tasks, evidence and Proof-of-Value instrumentation.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Financial Actions

Human-approved payment and collection intents with policy controls.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

MCP

Tool proposal and governed agent integration without implicit execution authority.
Open specification

The OpenAPI document is the machine-readable contract for endpoint definitions, required scopes, request schemas and response objects. Production boundaries remain scope-gated.

Usage & commercial evidence

LQM training, evaluation and promoted inference can be measured through the existing FinanceGPT usage control plane. Pricing is versioned from measured infrastructure-cost evidence rather than architecture assumptions, while experimental published LQMs remain uncallable until separately promoted.

GET /api/v2/lqm/commercial-model
GET /api/v2/lqm/usage-evidence
POST /api/v2/lqm/workflow-outcomes