# InferCrane

> Deploy and operate AI models across your infrastructure. Build or connect inference with durable operations, safe releases, and inspectable evidence.

## Docs

- [Deploy and operate AI models across your infrastructure.](https://docs.infercrane.com/index.md): Build or connect inference with durable operations, autoscaling, safe releases, and evidence you can inspect.
- [Why InferCrane](https://docs.infercrane.com/why-infercrane.md): The production inference problems InferCrane solves, the tools it complements, and when it is not the right choice.
- [Choose the right inference approach](https://docs.infercrane.com/compare.md): Compare InferCrane with managed inference platforms, model APIs, gateways, and Kubernetes-native stacks.
- [Quickstart](https://docs.infercrane.com/quickstart.md): Run InferCrane locally and send an OpenAI-compatible request in under five minutes.
- [Core concepts](https://docs.infercrane.com/concepts.md): The six objects you need to understand InferCrane's deployment and request model.
- [Architecture](https://docs.infercrane.com/architecture/system.md): InferCrane's persistent control plane, database-free inference path, and provider boundaries.
- [What you can build](https://docs.infercrane.com/showcase/index.md): Production inference patterns that start with one command and preserve evidence, ownership, and control.
- [Build your inference deployment](https://docs.infercrane.com/showcase/build-inference.md): Go from a model or immutable OCI workload to a durable, OpenAI-compatible endpoint.
- [Inference projects](https://docs.infercrane.com/features/inference-projects.md): Scaffold, validate, build, and deploy one reproducible inference workload from its own directory.
- [Connect the stack you already run](https://docs.infercrane.com/showcase/connect-existing.md): Get useful evidence from an existing vLLM, LiteLLM, or OpenAI-compatible endpoint without migrating it.
- [Stable endpoints and serving plans](https://docs.infercrane.com/features/endpoints.md): Keep application model names stable while changing concrete inference implementations.
- [DeploymentSpec](https://docs.infercrane.com/deployment-spec.md): Declaratively define model identity, runtime, provider, compute, scaling, and routing.
- [Inference protocols](https://docs.infercrane.com/features/protocols.md): Faithful OpenAI-compatible protocol proxying with explicit runtime capability gates.
- [Model artifacts](https://docs.infercrane.com/features/model-artifacts.md): Resolve mutable model references to immutable, evidence-bearing artifact identity.
- [Custom OCI workloads](https://docs.infercrane.com/features/custom-oci.md): Bring an immutable OpenAI-compatible inference image through the portable runtime contract.
- [Provisioning and readiness](https://docs.infercrane.com/features/provisioning.md): Follow capacity allocation, artifact preparation, runtime startup, readiness, and metrics.
- [Revisions and rollouts](https://docs.infercrane.com/features/revisions.md): Create immutable candidates, inspect evidence, promote safely, and roll back.
- [Deployment lifecycle](https://docs.infercrane.com/features/lifecycle.md): How durable operations survive disconnects, retries, restarts, updates, and deletion.
- [Observe operations](https://docs.infercrane.com/features/observe.md): Read one useful snapshot of a deployment or stable endpoint, then reconnect when work takes longer than a terminal session.
- [Terminal operations workspace](https://docs.infercrane.com/features/terminal-ui.md): Observe and safely operate deployments from a reconnectable terminal interface.
- [Operations console](https://docs.infercrane.com/features/dashboard.md): Use InferCrane's authenticated browser console without putting provider credentials or PostgreSQL in the frontend.
- [Autoscaling](https://docs.infercrane.com/features/autoscaling.md): Scale replicas with bounded decisions, durable evidence, and generation-safe drain.
- [Admission and async inference](https://docs.infercrane.com/features/admission-async.md): Bound overload before it reaches GPUs and run reconnectable inference jobs with encrypted persistence.
- [Release Guard](https://docs.infercrane.com/features/release-guard.md): Compare active and candidate evidence with deterministic, persisted promotion policy.
- [Semantic quality evidence](https://docs.infercrane.com/features/semantic-quality.md): Bind signed, aggregate task-quality evidence to exact revisions and enforce deterministic rollout thresholds.
- [Environment promotion](https://docs.infercrane.com/features/environment-promotion.md): Stage an immutable serving plan from development or staging into a destination endpoint without changing production traffic.
- [Governed external capacity](https://docs.infercrane.com/features/external-capacity.md): Explicit, budgeted health or queue overflow to an OpenAI-compatible external target.
- [Provider-native serverless](https://docs.infercrane.com/features/serverless.md): Preserve a stable endpoint while a qualified provider owns workers and scale-to-zero.
- [Inference telemetry](https://docs.infercrane.com/features/telemetry.md): Normalize latency, token, error, runtime, deployment, revision, and provider evidence.
- [Reproducible benchmarking](https://docs.infercrane.com/features/benchmarking.md): Run AIPerf workloads and persist exact model, runtime, hardware, and workload evidence.
- [Cold-start intelligence](https://docs.infercrane.com/features/cold-starts.md): Measure exposed cold-start boundaries without inventing provider-hidden timing.
- [Artifact cache and prewarming](https://docs.infercrane.com/features/artifact-cache.md): Record provider-neutral cache observations and durable prefetch intent without claiming an unverified cache hit.
- [Deterministic explanations](https://docs.infercrane.com/features/explanations.md): Explain degraded state, scaling, rollouts, and cold starts from persisted evidence.
- [Reject a bad release before it reaches production](https://docs.infercrane.com/showcase/safe-rollouts.md): Use bounded validation and deterministic Release Guard evidence to keep an active revision safe.
- [Operate through slow and uncertain capacity](https://docs.infercrane.com/showcase/capacity-evidence.md): Resume durable deployment work, explain cold starts, compare measured plans, and bound external overflow.
- [Recipes and Inference Lab](https://docs.infercrane.com/features/recipes-lab.md): Capture immutable serving evidence and compare measured candidates without hidden provisioning.
- [Replay and capacity intelligence](https://docs.infercrane.com/features/replay-capacity.md): Capture privacy-preserving workload shape and reason from observed capacity evidence.
- [Inference decisions](https://docs.infercrane.com/features/inference-decisions.md): Deterministic SLO policies, evidence-based recommendations, and explicit unknowns.
- [FinOps and advisory Autopilot](https://docs.infercrane.com/features/finops-autopilot.md): Explain sourced inference cost and create human-approved serving recommendations.
- [Inference Passports](https://docs.infercrane.com/features/inference-passports.md): Canonical, signed, reproducible evidence for an immutable inference revision.
- [Context Passport and Burst Guard](https://docs.infercrane.com/features/context-passport-burst.md): Durable logical session identity, delegated request survival, and budget-bounded overflow.
- [Integration model](https://docs.infercrane.com/integrations/overview.md): How providers, runtimes, artifacts, routers, metrics, and benchmark tools extend InferCrane without forking its lifecycle.
- [LiteLLM](https://docs.infercrane.com/integrations/litellm.md): Keep LiteLLM provider translation and credentials while InferCrane adds stable endpoint identity, lifecycle evidence, and release policy.
- [External agent sandboxes](https://docs.infercrane.com/integrations/sandboxes.md): Give an externally operated sandbox short-lived access to exactly one InferCrane endpoint without storing commands, files, prompts, or outputs.
- [Training artifact handoffs](https://docs.infercrane.com/integrations/training-artifacts.md): Attach signed, content-free lineage from an external training pipeline to one immutable InferCrane revision.
- [Python SDK](https://docs.infercrane.com/integrations/python.md): Submit and resume durable InferCrane operations from typed Python applications.
- [TypeScript SDK](https://docs.infercrane.com/integrations/typescript.md): Typed durable operations and cancellation-aware inference streams for Node.js.
- [Terraform provider](https://docs.infercrane.com/integrations/terraform.md): Manage logical InferCrane deployments without transferring provider-resource ownership to Terraform.
- [GitHub Actions](https://docs.infercrane.com/integrations/github-actions.md): Deterministic pull-request plans and protected InferCrane release checks.
- [MCP for coding agents](https://docs.infercrane.com/integrations/mcp.md): Give coding agents read-only access to persisted InferCrane evidence without granting deployment authority.
- [Compose gateways and agent sandboxes](https://docs.infercrane.com/showcase/gateways-and-sandboxes.md): Keep InferCrane focused on inference lifecycle while integrating gateways, external capacity, and sandbox runtimes through explicit boundaries.
- [vLLM](https://docs.infercrane.com/integrations/vllm.md): How InferCrane manages and routes requests to vLLM runtimes.
- [SGLang](https://docs.infercrane.com/integrations/sglang.md): The second Runtime Contract V1 engine profile.
- [RunPod](https://docs.infercrane.com/integrations/runpod.md): Configure the RunPod elastic and provider-native Serverless adapter.
- [AWS EC2 BYOC](https://docs.infercrane.com/integrations/aws-ec2.md): Configure the narrow, private-network AWS EC2 provider adapter.
- [GCP Compute BYOC](https://docs.infercrane.com/integrations/gcp-compute.md): Configure the private, identity-based GCP Compute provider adapter.
- [Provider profiles](https://docs.infercrane.com/integrations/provider-profiles.md): Understand executable, registered, and deferred infrastructure profiles.
- [Kubernetes](https://docs.infercrane.com/integrations/kubernetes.md): Run InferCrane-owned inference workloads in a namespace without adding another operator or scheduler.
- [AIPerf](https://docs.infercrane.com/integrations/aiperf.md): Run reproducible load generation without sending benchmark data elsewhere.
- [Hugging Face models](https://docs.infercrane.com/integrations/hugging-face.md): Resolve mutable Hugging Face model references to immutable artifacts.
- [Provider contract](https://docs.infercrane.com/architecture/provider-contract.md): The lifecycle, ownership, idempotency, inventory, and qualification contract for infrastructure adapters.
- [Runtime contract](https://docs.infercrane.com/architecture/runtime-contract.md): The capability, readiness, protocol, cancellation, and evidence contract for inference runtimes.
- [CLI reference](https://docs.infercrane.com/cli.md): Commands for configuring, deploying, observing, updating, and deleting InferCrane deployments.
- [HTTP control API](https://docs.infercrane.com/control-api.md): Authenticated v1 endpoints, roles, idempotency, and stable errors implemented by InferCrane.
- [Adopt an existing inference endpoint without transferring lifecycle ownership](https://docs.infercrane.com/api-reference/adoption/adopt-an-existing-inference-endpoint-without-transferring-lifecycle-ownership.md)
- [Explicitly promote an adopted endpoint to traffic-managed ownership](https://docs.infercrane.com/api-reference/adoption/explicitly-promote-an-adopted-endpoint-to-traffic-managed-ownership.md)
- [Inspect fresh and expired cache observations separately from prefetch intent](https://docs.infercrane.com/api-reference/model-artifacts/inspect-fresh-and-expired-cache-observations-separately-from-prefetch-intent.md)
- [Record bounded provider-native artifact cache evidence](https://docs.infercrane.com/api-reference/model-artifacts/record-bounded-provider-native-artifact-cache-evidence.md)
- [Request provider-adapter artifact prefetch](https://docs.infercrane.com/api-reference/model-artifacts/request-provider-adapter-artifact-prefetch.md)
- [Inspect an async job and its result when complete](https://docs.infercrane.com/api-reference/async-inference/inspect-an-async-job-and-its-result-when-complete.md)
- [Cancel queued or executing async inference](https://docs.infercrane.com/api-reference/async-inference/cancel-queued-or-executing-async-inference.md)
- [Submit one encrypted durable inference job](https://docs.infercrane.com/api-reference/async-inference/submit-one-encrypted-durable-inference-job.md)
- [List tenant audit events](https://docs.infercrane.com/api-reference/audit/list-tenant-audit-events.md)
- [Inspect an immutable advisory plan](https://docs.infercrane.com/api-reference/autopilot/inspect-an-immutable-advisory-plan.md)
- [Record human approval without mutating serving state](https://docs.infercrane.com/api-reference/autopilot/record-human-approval-without-mutating-serving-state.md)
- [Create an advisory plan from a persisted recommendation](https://docs.infercrane.com/api-reference/autopilot/create-an-advisory-plan-from-a-persisted-recommendation.md)
- [Summarize tenant-scoped observed capacity evidence](https://docs.infercrane.com/api-reference/capacity-intelligence/summarize-tenant-scoped-observed-capacity-evidence.md)
- [List private-console members for the authenticated organization](https://docs.infercrane.com/api-reference/identity/list-private-console-members-for-the-authenticated-organization.md)
- [Map a hosted identity and grant or revoke private-console access](https://docs.infercrane.com/api-reference/identity/map-a-hosted-identity-and-grant-or-revoke-private-console-access.md)
- [Resolve the authenticated console organization and entitlements](https://docs.infercrane.com/api-reference/identity/resolve-the-authenticated-console-organization-and-entitlements.md)
- [List service accounts for the authenticated organization](https://docs.infercrane.com/api-reference/identity/list-service-accounts-for-the-authenticated-organization.md)
- [Create a scoped service account](https://docs.infercrane.com/api-reference/identity/create-a-scoped-service-account.md)
- [Revoke a service account](https://docs.infercrane.com/api-reference/identity/revoke-a-service-account.md)
- [Rotate a service-account credential](https://docs.infercrane.com/api-reference/identity/rotate-a-service-account-credential.md)
- [Set tenant safety quotas](https://docs.infercrane.com/api-reference/identity/set-tenant-safety-quotas.md)
- [Create a tenant](https://docs.infercrane.com/api-reference/identity/create-a-tenant.md)
- [Inspect the authenticated principal](https://docs.infercrane.com/api-reference/identity/inspect-the-authenticated-principal.md)
- [Create bounded durable logical session identity](https://docs.infercrane.com/api-reference/context-passport/create-bounded-durable-logical-session-identity.md)
- [Inspect logical session identity and best-effort affinity](https://docs.infercrane.com/api-reference/context-passport/inspect-logical-session-identity-and-best-effort-affinity.md)
- [List logical deployments](https://docs.infercrane.com/api-reference/deployments/list-logical-deployments.md)
- [Create a provider-managed deployment](https://docs.infercrane.com/api-reference/deployments/create-a-provider-managed-deployment.md)
- [Apply a DeploymentSpec](https://docs.infercrane.com/api-reference/deployments/apply-a-deploymentspec.md)
- [Inspect deployment lifecycle state](https://docs.infercrane.com/api-reference/deployments/inspect-deployment-lifecycle-state.md)
- [Delete a logical deployment](https://docs.infercrane.com/api-reference/deployments/delete-a-logical-deployment.md)
- [List deployment events](https://docs.infercrane.com/api-reference/deployments/list-deployment-events.md)
- [List benchmark history](https://docs.infercrane.com/api-reference/benchmarks/list-benchmark-history.md)
- [Run and persist an AIPerf benchmark](https://docs.infercrane.com/api-reference/benchmarks/run-and-persist-an-aiperf-benchmark.md)
- [Persist a fresh budget-bounded overflow decision](https://docs.infercrane.com/api-reference/burst-guard/persist-a-fresh-budget-bounded-overflow-decision.md)
- [Inspect governed external fallback](https://docs.infercrane.com/api-reference/external-capacity/inspect-governed-external-fallback.md)
- [Configure governed external fallback](https://docs.infercrane.com/api-reference/external-capacity/configure-governed-external-fallback.md)
- [List persisted FinOps evidence](https://docs.infercrane.com/api-reference/finops/list-persisted-finops-evidence.md)
- [Persist a sourced cost report without invented savings](https://docs.infercrane.com/api-reference/finops/persist-a-sourced-cost-report-without-invented-savings.md)
- [List signed Inference Passports](https://docs.infercrane.com/api-reference/release-evidence/list-signed-inference-passports.md)
- [Issue a signed Inference Passport](https://docs.infercrane.com/api-reference/release-evidence/issue-a-signed-inference-passport.md)
- [List revision-bound semantic quality evidence without prompt or output content](https://docs.infercrane.com/api-reference/release-evidence/list-revision-bound-semantic-quality-evidence-without-prompt-or-output-content.md)
- [Attach signed aggregate semantic quality evidence to an immutable revision](https://docs.infercrane.com/api-reference/release-evidence/attach-signed-aggregate-semantic-quality-evidence-to-an-immutable-revision.md)
- [Capture an immutable evidence-backed recipe](https://docs.infercrane.com/api-reference/recipes/capture-an-immutable-evidence-backed-recipe.md)
- [Search immutable model recipes](https://docs.infercrane.com/api-reference/recipes/search-immutable-model-recipes.md)
- [Get one immutable model recipe](https://docs.infercrane.com/api-reference/recipes/get-one-immutable-model-recipe.md)
- [List persisted recommendation history](https://docs.infercrane.com/api-reference/inference-decisions/list-persisted-recommendation-history.md)
- [Evaluate and persist a recommendation](https://docs.infercrane.com/api-reference/inference-decisions/evaluate-and-persist-a-recommendation.md)
- [Inspect deterministic SLO policy](https://docs.infercrane.com/api-reference/inference-decisions/inspect-deterministic-slo-policy.md)
- [Set deterministic SLO policy](https://docs.infercrane.com/api-reference/inference-decisions/set-deterministic-slo-policy.md)
- [Delete deterministic SLO policy](https://docs.infercrane.com/api-reference/inference-decisions/delete-deterministic-slo-policy.md)
- [Inspect the persisted guard policy](https://docs.infercrane.com/api-reference/release-guard/inspect-the-persisted-guard-policy.md)
- [Set the persisted guard policy](https://docs.infercrane.com/api-reference/release-guard/set-the-persisted-guard-policy.md)
- [Evaluate active and candidate revisions](https://docs.infercrane.com/api-reference/release-guard/evaluate-active-and-candidate-revisions.md)
- [Capture a privacy-preserving production workload shape](https://docs.infercrane.com/api-reference/replay/capture-a-privacy-preserving-production-workload-shape.md)
- [Inspect a persisted workload-shape trace](https://docs.infercrane.com/api-reference/replay/inspect-a-persisted-workload-shape-trace.md)
- [List immutable deployment revisions](https://docs.infercrane.com/api-reference/revisions/list-immutable-deployment-revisions.md)
- [Roll back to a prior revision](https://docs.infercrane.com/api-reference/revisions/roll-back-to-a-prior-revision.md)
- [Create a candidate revision](https://docs.infercrane.com/api-reference/revisions/create-a-candidate-revision.md)
- [Promote a validated candidate](https://docs.infercrane.com/api-reference/revisions/promote-a-validated-candidate.md)
- [Provision candidate capacity](https://docs.infercrane.com/api-reference/revisions/provision-candidate-capacity.md)
- [Reject a candidate revision](https://docs.infercrane.com/api-reference/revisions/reject-a-candidate-revision.md)
- [Set the active route target](https://docs.infercrane.com/api-reference/routing/set-the-active-route-target.md)
- [List persisted scaling decisions](https://docs.infercrane.com/api-reference/scaling/list-persisted-scaling-decisions.md)
- [List content-free external training provenance for deployment revisions](https://docs.infercrane.com/api-reference/training-lineage/list-content-free-external-training-provenance-for-deployment-revisions.md)
- [Verify and attach a signed immutable checkpoint handoff](https://docs.infercrane.com/api-reference/training-lineage/verify-and-attach-a-signed-immutable-checkpoint-handoff.md)
- [Run control-plane diagnostics](https://docs.infercrane.com/api-reference/system/run-control-plane-diagnostics.md)
- [Inspect registered integration capabilities](https://docs.infercrane.com/api-reference/system/inspect-registered-integration-capabilities.md)
- [List live control-plane instances and protocol compatibility](https://docs.infercrane.com/api-reference/system/list-live-control-plane-instances-and-protocol-compatibility.md)
- [List stable application endpoints](https://docs.infercrane.com/api-reference/endpoints/list-stable-application-endpoints.md)
- [Create a stable application endpoint](https://docs.infercrane.com/api-reference/endpoints/create-a-stable-application-endpoint.md)
- [Inspect an endpoint and its serving plans](https://docs.infercrane.com/api-reference/endpoints/inspect-an-endpoint-and-its-serving-plans.md)
- [Delete an endpoint without deleting its bound deployment](https://docs.infercrane.com/api-reference/endpoints/delete-an-endpoint-without-deleting-its-bound-deployment.md)
- [Attach a backend binding to an endpoint](https://docs.infercrane.com/api-reference/endpoints/attach-a-backend-binding-to-an-endpoint.md)
- [Create and stage an immutable serving plan](https://docs.infercrane.com/api-reference/endpoints/create-and-stage-an-immutable-serving-plan.md)
- [Promote a serving plan for future requests](https://docs.infercrane.com/api-reference/endpoints/promote-a-serving-plan-for-future-requests.md)
- [Stage a candidate serving plan](https://docs.infercrane.com/api-reference/endpoints/stage-a-candidate-serving-plan.md)
- [Evaluate active and candidate serving plans](https://docs.infercrane.com/api-reference/endpoints/evaluate-active-and-candidate-serving-plans.md)
- [List endpoint Release Guard decisions](https://docs.infercrane.com/api-reference/endpoints/list-endpoint-release-guard-decisions.md)
- [Inspect endpoint Release Guard policy](https://docs.infercrane.com/api-reference/endpoints/inspect-endpoint-release-guard-policy.md)
- [Set endpoint Release Guard policy](https://docs.infercrane.com/api-reference/endpoints/set-endpoint-release-guard-policy.md)
- [Stage one environment endpoint's immutable plan as another endpoint's candidate](https://docs.infercrane.com/api-reference/endpoints/stage-one-environment-endpoints-immutable-plan-as-another-endpoints-candidate.md)
- [List endpoint environments](https://docs.infercrane.com/api-reference/endpoints/list-endpoint-environments.md)
- [Create an endpoint environment](https://docs.infercrane.com/api-reference/endpoints/create-an-endpoint-environment.md)
- [List logical model identities](https://docs.infercrane.com/api-reference/endpoints/list-logical-model-identities.md)
- [Create a logical model identity](https://docs.infercrane.com/api-reference/endpoints/create-a-logical-model-identity.md)
- [Inspect bounded endpoint admission policy](https://docs.infercrane.com/api-reference/admission/inspect-bounded-endpoint-admission-policy.md)
- [Set bounded endpoint admission policy](https://docs.infercrane.com/api-reference/admission/set-bounded-endpoint-admission-policy.md)
- [List signed webhook alert policies](https://docs.infercrane.com/api-reference/alerts/list-signed-webhook-alert-policies.md)
- [Create a signed webhook alert policy](https://docs.infercrane.com/api-reference/alerts/create-a-signed-webhook-alert-policy.md)
- [Evaluate deterministic findings and deliver eligible alerts](https://docs.infercrane.com/api-reference/alerts/evaluate-deterministic-findings-and-deliver-eligible-alerts.md)
- [Persist deterministic endpoint findings](https://docs.infercrane.com/api-reference/diagnostics/persist-deterministic-endpoint-findings.md)
- [Inspect persisted request metadata without content](https://docs.infercrane.com/api-reference/diagnostics/inspect-persisted-request-metadata-without-content.md)
- [Read bounded normalized inference and lifecycle evidence](https://docs.infercrane.com/api-reference/monitoring/read-bounded-normalized-inference-and-lifecycle-evidence.md)
- [Compare persisted serving evidence](https://docs.infercrane.com/api-reference/inference-lab/compare-persisted-serving-evidence.md)
- [List durable operations for the authenticated organization](https://docs.infercrane.com/api-reference/operations/list-durable-operations-for-the-authenticated-organization.md)
- [Inspect a durable operation](https://docs.infercrane.com/api-reference/operations/inspect-a-durable-operation.md)
- [Request cooperative operation cancellation](https://docs.infercrane.com/api-reference/operations/request-cooperative-operation-cancellation.md)
- [List durable operation events](https://docs.infercrane.com/api-reference/operations/list-durable-operation-events.md)
- [List orphaned provider resources](https://docs.infercrane.com/api-reference/infrastructure/list-orphaned-provider-resources.md)
- [List external sandbox references without command, file, prompt, or output content](https://docs.infercrane.com/api-reference/sandboxes/list-external-sandbox-references-without-command-file-prompt-or-output-content.md)
- [Register an externally owned sandbox and issue an expiring endpoint-scoped credential](https://docs.infercrane.com/api-reference/sandboxes/register-an-externally-owned-sandbox-and-issue-an-expiring-endpoint-scoped-credential.md)
- [Revoke InferCrane access without mutating the external sandbox](https://docs.infercrane.com/api-reference/sandboxes/revoke-infercrane-access-without-mutating-the-external-sandbox.md)
- [Rotate the expiring endpoint-scoped credential for an active sandbox reference](https://docs.infercrane.com/api-reference/sandboxes/rotate-the-expiring-endpoint-scoped-credential-for-an-active-sandbox-reference.md)
- [List secret references](https://docs.infercrane.com/api-reference/secrets/list-secret-references.md)
- [Create a reference to an external secret](https://docs.infercrane.com/api-reference/secrets/create-a-reference-to-an-external-secret.md)
- [Delete a secret reference](https://docs.infercrane.com/api-reference/secrets/delete-a-secret-reference.md)
- [List inference targets](https://docs.infercrane.com/api-reference/targets/list-inference-targets.md)
- [Register an inference target](https://docs.infercrane.com/api-reference/targets/register-an-inference-target.md)
- [Create a buffered or streaming chat completion](https://docs.infercrane.com/api-reference/inference/create-a-buffered-or-streaming-chat-completion.md): When stream=true, the response is an SSE sequence of data JSON events terminated by data: [DONE]. InferCrane never replays a partially transmitted stream.
- [Proxy a capability-qualified OpenAI-compatible request](https://docs.infercrane.com/api-reference/inference/proxy-a-capability-qualified-openai-compatible-request.md): The selected endpoint must explicitly qualify this protocol. InferCrane rewrites only the logical model identity and otherwise preserves the request and response.
- [Proxy a capability-qualified OpenAI-compatible request](https://docs.infercrane.com/api-reference/inference/proxy-a-capability-qualified-openai-compatible-request-1.md): The selected endpoint must explicitly qualify this protocol. InferCrane rewrites only the logical model identity and otherwise preserves the request and response.
- [Proxy a capability-qualified OpenAI-compatible request](https://docs.infercrane.com/api-reference/inference/proxy-a-capability-qualified-openai-compatible-request-2.md): The selected endpoint must explicitly qualify this protocol. InferCrane rewrites only the logical model identity and otherwise preserves the request and response.
- [List logical inference models](https://docs.infercrane.com/api-reference/inference/list-logical-inference-models.md)
- [Proxy a capability-qualified OpenAI-compatible request](https://docs.infercrane.com/api-reference/inference/proxy-a-capability-qualified-openai-compatible-request-3.md): The selected endpoint must explicitly qualify this protocol. InferCrane rewrites only the logical model identity and otherwise preserves the request and response.
- [Provider setup](https://docs.infercrane.com/provider-setup.md): Configure credentials and prerequisites for qualified infrastructure adapters.
- [PostgreSQL backup and restore](https://docs.infercrane.com/runbooks/backup-restore.md): Back up control-plane state, restore into a safe target, and reconcile external resources.
- [Upgrade and compatibility](https://docs.infercrane.com/upgrade.md): Version guarantees, database migration safety, rolling-upgrade boundaries, and rollback policy.
- [Development and testing](https://docs.infercrane.com/development.md): Fast local feedback, provider contracts, Docker qualification, and paid acceptance boundaries.
- [Troubleshooting](https://docs.infercrane.com/troubleshooting.md): Diagnose configuration, provisioning, readiness, routing, rollout, and cleanup failures.
- [Frequently asked questions](https://docs.infercrane.com/faq.md): Product scope, compatibility, data handling, serverless, sessions, and pricing answers.
- [Capability status](https://docs.infercrane.com/project-status.md): The authoritative implementation and qualification state of every InferCrane capability.
- [Roadmap](https://docs.infercrane.com/roadmap.md): InferCrane's product direction, priorities, and explicit non-goals.
- [Product qualification](https://docs.infercrane.com/testing/product-qualification.md): Run InferCrane's resumable, commit-bound local and real-infrastructure evidence system.
- [Reliability test environment](https://docs.infercrane.com/testing/test-environment.md): InferCrane's long-term local, scheduled, simulated, and real-infrastructure qualification strategy.
- [Security](https://docs.infercrane.com/security.md): Authentication, tenant isolation, secret references, network boundaries, and secure deployment guidance.
- [Compatibility and qualification](https://docs.infercrane.com/compatibility.md): How InferCrane separates registered adapters, local conformance, and real-infrastructure support evidence.

## OpenAPI Specs

- [openapi](https://docs.infercrane.com/openapi.json)