generated: '2026-09-02' method: searched source: https://standardcompute.com/status also: - https://standardcompute.com/changelog - https://standardcompute.com/fair-use - https://standardcompute.com/pricing versioning: scheme: path-prefix current: v1 base: https://api.stdcmpt.com/v1 spec_version: '0.1.0' spec_version_note: >- The published OpenAPI carries info.version 0.1.0 and info.title "FastAPI" — framework defaults that have never been set to a real product version, so the contract cannot be used to tell one release from another. policy_published: false note: >- No versioning policy page exists. /v1 has been stable since the November 2025 launch and no v2 has been announced. model_lifecycle: managed_by_provider: true statement: >- Callers set model to "standardcompute" and the router selects the underlying model per request. The provider frames this as a lifecycle FEATURE: "you're never stuck on a deprecated model". The corollary an integrator should weigh is that the caller cannot pin a model version, so upstream model changes arrive without a client-side change and without a client-side veto. source: https://standardcompute.com/llms-full.txt observed_pool: probed: GET https://api.stdcmpt.com/v1/models (2026-09-02, HTTP 200, unauthenticated) ids: - StandardCompute - anthropic/claude-standardcompute - anthropic/claude-fable-5 - anthropic/claude-opus-5 - anthropic/claude-opus-5-fast - anthropic/claude-sonnet-5 deprecation: policy_published: false sunset_header: false deprecation_header: false rfc8594: false deprecated_operations: [] change_notice: >- The fair-use policy commits to 14 days' notice of MATERIAL CHANGES to that policy. That is a terms-change notice, not an API deprecation policy, so no Deprecation pointer is wired — recording the distinction rather than crediting a policy the provider has not published. source: https://standardcompute.com/fair-use plan_grandfathering: >- Documented and honoured in practice: the 2026-08-29 plan-ladder change renamed the $249 "Max" tier to "Pro" at the same price and states all grandfathered plans continue unchanged. status_page: url: https://standardcompute.com/status vendor: first-party machine_readable: https://standardcompute.com/api/status machine_readable_status: 200 machine_readable_probed: '2026-09-02' format: >- JSON — {gateway:{status,latency_ms}, services:[{name,status}], timestamp}. Not a standard status schema, but genuinely machine consumable, which is rare at this company size. live_checks: >- The page is not hand-set: loading it fires real requests at the production endpoints and renders what came back, including gateway latency. components: - API Gateway - /v1/chat/completions - /v1/completions - /v1/responses - Authentication - Dashboard self_verification: >- Documented: "curl -i https://api.stdcmpt.com/v1/models" — the models endpoint answers unauthenticated, so any caller can independently confirm the gateway. sla: published: true target: 99.9% monthly uptime for the core API at https://api.stdcmpt.com/v1 scope: gateway, authentication, key validation, request routing exclusions: upstream model provider outages (mitigated by cross-provider failover) contractual: >- Stated as a target on the public status page. A contractual uptime SLA is offered only on the custom Business tier. source: https://standardcompute.com/status maintenance: announced_in_advance: true typical_window: weekdays 06:00-08:00 UTC source: https://standardcompute.com/status incidents: post_incident_reviews: >- Committed for significant incidents — "what happened, the timeline, and what we changed" — published on the status page. No archived incident was visible at probe time. maturity: first_public: '2025-04-07 (closed alpha)' open_beta: '2025-07-14' general_availability: '2025-11-18' source: https://standardcompute.com/changelog