GLM-5.3 API relay at $0.6528: one Quanta balance for bills

01 · How can several vendor bills become one Quanta balance?

When a team runs production workloads across several model providers, the administrative cost is rarely the API call itself. It is the stack of separate accounts, separate balances, separate invoices and separate payment terms that comes with them. Each provider adds another reconciliation task, and each top-up becomes another line for finance to track.

Quanta gives those workloads a single place to land. Instead of managing a balance with one vendor, a second balance with another and a third invoice somewhere else, a team funds one prepaid Quanta balance and draws on it across the models it uses. Quanta is built for teams and heavy, sustained workloads, and the point of that design is straightforward: fewer billing objects to manage, and a clearer view of what each call actually cost.

There is no monthly fee and no minimum spend. Quanta is pay-as-you-go only, so the balance reflects real usage rather than a commitment to a tier. Every call is metered on real tokens, and the dashboard lists the tokens and the charge of every individual call, which turns a consolidated bill back into a readable set of line items.

The practical result is that a team can route different tasks to different models while keeping settlement in one currency and one account. For example, a team might send long-context work to a Claude model, use GPT-6.1 Sol for routine automation, and place lighter classification jobs on Gemini 2.5 Flash. On Quanta, those choices do not fragment the billing picture.

02 · What does Quanta actually charge for each call?

Quanta prices input, output, cache writes and cache reads separately. That separation matters for teams with sustained traffic, because a workload that leans heavily on cached context has a different cost profile from one that resends the same material every time. The dashboard shows the token counts and the charge for every call, so a team can see which pattern is driving the bill rather than guessing from a monthly total.

GLM-5.3, for instance, is listed at $0.6528 per 1M input tokens, $1.97 per 1M output tokens and $0.1667 per 1M cache read tokens. Qwen3.7 Plus is listed at $0.1667 per 1M input tokens and $0.4861 per 1M output tokens. Claude Sonnet 5.5 is listed at $1.25 per 1M input tokens and $6.28 per 1M output tokens, with $1.57 per 1M cache writes and $0.0917 per 1M cache reads. These are Quanta prices for the listed models, and each is metered against the tokens actually used.

To make the arithmetic concrete, here are examples. Using a 3:1 input-to-output ratio with no cache hits, GPT-6.1 Sol works out to approximately $4.10 for 10 million tokens a month and $40.97 for 100 million. Under the same example conditions, Gemini 2.5 Flash comes to approximately $6.49 for 10 million tokens a month and $64.93 for 100 million. Claude Sonnet 5.5 comes to approximately $25.07 for 10 million tokens a month and $250.69 for 100 million. These are illustrations, not quotes, but they show how a single balance can absorb several price points without several separate billing relationships.

03 · Can one key cover different protocols and different models?

Yes. Quanta supports the OpenAI-compatible endpoint, Anthropic native and Gemini native, and one key and one balance work across all three. That is the practical core of bill consolidation: the integration surface can differ by model family while the account and the money stay in one place.

Client pairing is worth stating precisely, because it affects how a team migrates existing traffic. Claude models work only from the official Claude Code and Claude Desktop clients, with Claude Code 2.1.285 or newer. Gemini models use the native protocol only, through Gemini CLI. Codex, Cursor, Cherry Studio, Cline, OpenCode, OpenAI SDKs and cURL pair only with GPT, DeepSeek and the other non-Claude, non-Gemini models available on Quanta, including GLM-5.3, Qwen3.8 Max, Kimi K3, Hunyuan 4 and Grok 4.7.

So a team can, for example, keep an OpenAI-compatible setup for GPT-5.6 Sol or GLM-5.3, run Gemini 3.1 Pro through Gemini CLI, and use Claude Opus 5.5 from Claude Code or Claude Desktop. Three access patterns, one balance, one dashboard. The API base address for Quanta is the same regardless of which supported protocol a request uses.

On data handling, Quanta keeps the text users send for 30 days for abuse handling and billing disputes, then deletes it. Model answers are not stored. For a finance team reviewing a consolidated bill, that means the record kept for dispute resolution is defined and time-limited rather than open-ended.

04 · Will a consolidated balance survive heavy, sustained traffic?

Consolidation only helps if the relay can hold up under load, so Quanta publishes its limits and enforces them. Each key is limited to 120 requests per minute. Each account can have 8 requests in flight at once. A single request body is capped at 25 MB, and an account can hold 20 live keys.

Streaming requests have no total-time cap; a stream ends only after 300 seconds of upstream silence. That distinction is important for long-running generation jobs. A stream that is actively producing tokens is not cut off because it has run for a certain number of minutes, which suits pipelines where responses are consumed incrementally rather than all at once.

Because the limits are stated in plain numbers, a team can size its concurrency before it commits funds. Eight requests in flight per account is a known ceiling rather than an unknown one, and 120 requests per minute per key gives a clear basis for planning how many keys a given workload needs. Quanta also supports up to 20 live keys per account, so separate keys can carry separate workloads while still drawing on the same balance.

05 · How does company payment and invoicing work with Quanta?

Company customers can pay by bank transfer from a company account. Invoices and advance quotes are arranged through the Contact page or the Messages page of the dashboard. There is no support email address to hunt for, because the conversation stays inside the account the team already uses.

Top-ups are available at $1.39, $6.94, $13.89, $27.78, $69.44 and $138.89, so a team can start small and scale its balance as usage becomes predictable. Unused balance can be refunded within 15 days of the top-up. That refund window gives finance a defined period in which a balance that was topped up but not yet consumed can be returned, which is a useful answer when consolidating spend across several vendors.

There is also a referral reward structure: 19% of a referred account's top-ups, and 5% one level further. Rewards can be withdrawn from $27.78 or moved into the balance. For teams already running large token volumes, moving rewards into the balance means they contribute directly to the same pool that pays for production traffic.

06 · What is the first step to folding bills into one balance?

Start with the trial credit of $0.50 on sign-up and run a representative workload against it. The point is not to treat the trial as a permanent cost, but to check the shape of the numbers: which models the team actually calls, how many requests per minute each key needs, how much of the traffic benefits from cache reads, and what the dashboard shows for individual calls.

Then top up the balance and move the recurring traffic. Because Quanta charges on real tokens with no monthly fee and no minimum spend, the balance tracks usage rather than a subscription. And because concurrency, rate limits, request size and key counts are published, the team can plan against actual constraints instead of negotiating them one vendor at a time.

For teams tired of reconciling several vendor bills, the offer from Quanta is direct: one prepaid balance, one dashboard, token-level visibility on every call, and a single invoice relationship arranged through the pages you already have access to. Open an account, fund it once, and see where the rest of the billing stack disappears.