# ThinkMaaS

> ThinkMaaS is an LLM aggregation platform (Model-as-a-Service) by ThinkAlike. A unified gateway aggregates GLM, DeepSeek, Qwen, GPT, Claude and Gemini behind one OpenAI-compatible API, with smart multi-channel routing, automatic failover, token quotas, usage billing and a cost dashboard — powering every business system above it.

- Product page: https://thinkalike.com.cn/en/thinkmaas.html
- Markdown: https://thinkalike.com.cn/en/thinkmaas.md
- Free trial: https://maas.thinkalike.com.cn

## What it is

Teams often apply for model keys ad hoc — scattered secrets, messy bills, high switching cost. ThinkMaaS adds one gateway layer above all upstream models: business systems use a single endpoint and one auth scheme while the gateway handles routing, orchestration, metering and governance, converging scattered model calls into one observable foundation.

## Core capabilities

- **Unified model gateway** — one OpenAI-compatible endpoint for all models; connect, aggregate and scale channels with zero business-code changes.
- **Smart multi-channel routing** — attach multiple channels per model and route by weight, price and health with automatic load balancing.
- **Automatic failover** — when a channel is rate-limited, slow or errors out, requests switch seamlessly to backups with little user impact.
- **Tokens & quotas** — issue per-team / app / environment tokens with budget, rate and allowed-model limits for clear permission boundaries.
- **Usage billing & cost dashboard** — per-token metering, multi-currency conversion and normalized cost make every dollar of AI spend attributable and budgetable.
- **Private deployment** — Docker delivery runs inside government/enterprise intranets offline; model traffic and keys never leave your domain, pairing with RMS for domestic GPU monitoring.

## Supported models

Zhipu GLM-4.6 / GLM-5, DeepSeek-V3 / R1, Qwen, OpenAI GPT series, Anthropic Claude, Google Gemini, ERNIE Bot, Kimi/Moonshot, and locally deployed open-source models — the list keeps expanding.

## Free-quota campaign: ThinkMaaS Free Token Hunter

Alongside paid usage, ThinkMaaS runs a zero-friction campaign page: register an account (email verification is the rollout precondition), claim a trial grant and top it up with a daily check-in, then call a pool of models that are genuinely free right now through the same OpenAI-compatible endpoint `https://maas.thinkalike.com.cn/v1`. The free capacity is discovered and tracked by ThinSpider, quota is claimed on our own verified vendor accounts, and everything is metered through the gateway.

- Campaign page: https://thinkalike.com.cn/en/free-token.html | 中文: https://thinkalike.com.cn/free-token.html | Markdown: https://thinkalike.com.cn/en/free-token.md
- Verifiable figures (observed 2026-10-05): 17 of the 466 entries in the OpenRouter model catalog have ids ending in `:free` (https://openrouter.ai/api/v1/models ); free-model requests are capped at 20 requests per minute, 50 requests per day under 10 USD cumulative spend and 1000 per day above it (https://openrouter.ai/docs/api-reference/limits ); platform fields `register_enabled=true`, `email_verification=true`, `checkin_enabled=false`, `quota_per_unit=500000` (https://maas.thinkalike.com.cn/api/status , i.e. registration and email verification are live while the grant and daily check-in are not yet opened).
- Free-quota users can only call models inside the free-pool group; paid channels (including existing ThinKM gateway channels) are invisible and unusable to them, and existing channels are unaffected.
- Quota follows upstream policy: no unlimited quota, no permanent free tier, no availability or SLA claim. Sources unreachable from our runtime host (Gemini / Mistral / ModelScope / Groq) neither enter the runtime pool nor get advertised.

## Use cases

Enterprises, research institutes and academia that need to unify multi-model calls, control AI cost and guarantee high availability and compliance; a shared model foundation for digital employees, marketing, monitoring and training systems.

## FAQ

**What is ThinkMaaS?**
An LLM aggregation platform (Model-as-a-Service) by ThinkAlike that unifies GLM, DeepSeek, Qwen, GPT, Claude and Gemini behind one gateway to govern your organization's AI usage, tokens and cost.

**How does it reduce AI cost?**
Smart multi-channel routing and automatic failover pick the best-value channel at equal quality; token quotas and usage billing make cost attributable and budgetable; new users enjoy 50% off the GLM family plus first top-up bonuses.

**Does it support private or air-gapped deployment?**
Yes. Docker-based delivery runs inside government/enterprise intranets offline; model traffic and key data never leave your domain, and it pairs with RMS to monitor domestic GPU compute.

**How do I try ThinkMaaS or deploy it on-premises?**
Visit https://maas.thinkalike.com.cn for an online trial, or contact us for an offline, on-premises deployment plan.