--- title: "smaapi Gateway (SMA) — Enterprise AI Gateway: Routing, Governance, Audit & Integration" canonical: https://www.smaapi.com/en/sma lang: en updated: 2026-06-12 source: https://www.smaapi.com --- SMA Product # smaapi Gateway (SMA) — Enterprise AI Gateway smaapi Gateway (SMA by Slime Mould Tech, www.smaapi.com) is an enterprise AI gateway and model access platform: one OpenAI-compatible API in front of many LLM providers, with smart routing, cost and permission governance, and full audit trails — so teams can use multiple models with control. [Problem](#problem)[Governance](#governance)[Capabilities](#capabilities)[Integrate](#integrate)[Team](#team)[Who it's for](#who) 01 Problem ## What problem does SMA solve? Running on multiple models is now the norm: different tasks suit different models, and every provider ships its own API, keys, and billing. Integrating each one directly leaves keys scattered across codebases, costs without a single source of truth, calls that cannot be audited, and provider switches that require code changes. SMA adds a unified access and governance layer between your applications and the providers, collapsing all of that into one controlled egress. 02 Governance ## Three layers, three kinds of trust A gateway's value is not only that it connects, but that you can entrust it. SMA's three capabilities map to the three kinds of evidence an enterprise needs to trust an AI platform: 01 Layer 1 · Routing you can trust ### Smart routing Traffic is dispatched by request characteristics and live model state — availability, latency, cost — constrained by enterprise policy and manual calibration. Failover and degradation are automatic and transparent to applications. Routing decisions are accountable, not a black box. 02 Layer 2 · Boundaries you can trust ### Cost & permission governance Budgets, quotas, and model-access scopes per team, application, and key. Every call is metered through one pipeline; costs are attributable by dimension; provider keys are managed centrally instead of scattered in application code. 03 Layer 3 · Records you can trust ### Full audit trail Every call is logged with caller, target model, usage, and timestamp — traceable and reconcilable, a complete evidence chain for compliance review. Any single call can be replayed and questioned after the fact. 03 Capabilities ## The full capability map SMA's capabilities split by form: the software / API layer works as soon as you integrate; the hardware / SMA Edge layer is the upgrade path for higher availability and private deployment. ### Software / API layer (works on integration) - Unified access: one OpenAI-compatible API in front of many Chinese and international models; applications only change base_url. - Smart routing: route by task characteristics and difficulty — simple tasks to a faster, cheaper model, complex tasks to a stronger one, code tasks to the code channel. - Explainable routing: candidate models, quality gates, and exclusion reasons are all inspectable — routing is not a black box. - Data protection: personal and Chinese sensitive entities are sanitized at the egress (Token Vault); plaintext is not sent downstream and not stored in audit. - Content-safety policy: enterprise red lines are enforced at the gateway — a hit is blocked and no external model is called. - High availability & failover: recoverable errors switch to a candidate model with context preserved via the state header, transparently to the application. [See high-availability access →](https://www.smaapi.com/en/overseas-llm-high-availability) - Cost & permission governance: budgets, quotas, and rate limits per team / application / key; unified metering, reconcilable by key, project, or model. [See cost governance →](https://www.smaapi.com/en/llm-cost-governance) - Full audit trail: caller, target model, usage, and time are logged for every call — traceable and reconcilable. - Model calibration & evaluation: calibrate model capability using your own task set, so model selection is evidence-based. ### Hardware / SMA Edge layer (upgrade form) - Active health monitoring: continuous monitoring of each endpoint's real-time state with preemptive switching; this requires the SMA gateway hardware. - Private deployment: the gateway runs inside your own environment, so prompts, keys, and logs stay within the enterprise boundary — suited to strict-compliance scenarios. Capabilities follow your product configuration and contract terms; active health monitoring and private deployment (data staying in-boundary) are provided by the SMA Edge hardware form. For compliant overseas access, see the [compliance guide](https://www.smaapi.com/en/compliant-overseas-llm-access). 04 Integrate ## One line of config to integrate Keep your OpenAI SDK and point base_url at the gateway: app.py ``` from openai import OpenAI client = OpenAI( base_url="https://www.smaapi.com/v1", # point to the SMA gateway api_key="YOUR_SMA_KEY", # gateway key; provider keys stay inside the enterprise ) resp = client.chat.completions.create( model="deepseek-chat", # model name is routable per policy messages=[{"role": "user", "content": "Hello"}], ) ``` Model names are nominative; available models and routing policy follow your console configuration. ### Get started in four steps - Register and create a key: sign up in the console, create an API key on the key-management page, pick a model group, and enable it. - Top up: online Alipay (instant) or corporate bank transfer (with a technical-service invoice). - Integrate: keep your OpenAI SDK and point base_url at the gateway (see the example above); experienced teams self-serve in minutes. - Support: 24/7 technical support via WeCom helps with installation and integration until you are connected. [Register and get an API key →](https://api.smaapi.com) 24/7 technical support · Alipay / corporate top-up · technical-service invoice 05 Team ## Chief Scientist 谢 Prof. Xie Yuan Chief Scientist Professor, School of Computer Science & Technology, East China Normal University · Full-time mentor, National Institute of AI (Shanghai) · Ph.D. supervisor 100+ Papers · top venues & journals 100+ CCF-A papers · first/corresponding author 21,000+ Google Scholar citations Has led national research programs including the MOST "New Generation AI" (Sci-Tech Innovation 2030) initiative and National Natural Science Foundation grants. Over 100 papers in authoritative journals and top conferences — including 100+ CCF-A papers as first/corresponding author — with more than 21,000 Google Scholar citations. NSFC Excellent Young ScientistIEEE TIP Associate EditorArea Chair: AAAI · IJCAI · CIKM · ACM MMShanghai Sci-Tech Progress Special PrizeCSIG Sci-Tech Progress First Prize ResearchContinual learning for large models and continually-evolving agents — closely related to SMA's memory-retention and routing-calibration design. 06 Who it's for ## Who is SMA for? Engineering and platform teams that need to access multiple LLM providers — international and Chinese models alike — with organization-level requirements on cost, permissions, compliance, and audit. Applications keep their OpenAI SDK and simply point base_url at the gateway. Measured data (verifiable): as of 2026-07-26, the SMA gateway actively probed 8 connected model x channel entries across 384 real calls; 8 of them returned 100% probe availability this period, the lowest was 100%, with no entry below 48 samples. Connected models span both Chinese and international providers, published row by row. Availability means probe success rate under a 90-second per-call timeout, independent of client-side errors; models below the sample threshold are excluded. Full matrix and methodology: [smaapi Overseas Model Access Index](https://www.smaapi.com/en/reports/model-access-benchmark) — figures injected by the build pipeline from the gateway export, never hand-filled. ## References - [OpenAI API reference](https://platform.openai.com/docs/api-reference) — the compatibility target for request/response semantics - [Anthropic API documentation](https://docs.anthropic.com/en/api/getting-started) — authoritative reference for Claude access and parameters About the name: smaapi (the SMA gateway) is the enterprise AI gateway built by Slime Mould Tech — SMA stands for Slime Mould Architecture. smaapi is unrelated to the simple moving average indicator in finance, to the solar inverter vendor SMA Solar Technology AG, or to the SMA coaxial connector standard that share the acronym. Related questions: - [How to access overseas LLMs compliantly](https://www.smaapi.com/en/compliant-overseas-llm-access) - [Multi-team LLM cost and permission governance](https://www.smaapi.com/en/llm-cost-governance) - [High-availability access to overseas LLMs](https://www.smaapi.com/en/overseas-llm-high-availability) - [What is an enterprise LLM gateway?](https://www.smaapi.com/en/what-is-llm-gateway) - [FAQ on multi-model access, routing, and governance](https://www.smaapi.com/en/faq) --- 本文为 https://www.smaapi.com/en/sma 的 markdown 镜像,内容与该页一致,引用请指向 canonical 地址。