Claude Fable 5 Pricing on Quanta: $4.35 Setup for Teams

01 · What Claude Fable 5 costs on Quanta

Quanta is built for teams running heavy, sustained workloads, and Claude Fable 5 is priced in the open. On Quanta, Claude Fable 5 (claude-fable-5) is listed at $4.35 per 1M tokens for input, $21.74 per 1M tokens for output, $5.35 per 1M tokens for cache writes, and $0.3333 per 1M tokens for cache reads. These are approximate USD figures; charges are made in CNY. Quanta meters every call on real tokens, and the dashboard lists the tokens and the charge for each call, so a team can reconcile spend line by line instead of estimating it after the fact.

Quanta is pay-as-you-go. There is no monthly fee, no minimum spend, and no plan tiers. You add prepaid balance and draw it down as work happens.

• › Input: $4.35 per 1M tokens
• › Output: $21.74 per 1M tokens
• › Cache write: $5.35 per 1M tokens
• › Cache read: $0.3333 per 1M tokens

Cache reads are listed separately from cache writes, which matters for teams that keep large context warm across a long session. Because Quanta records each of the four token classes on the call itself, the cost of a long-running job is visible while it runs. Nothing about the invoice has to be reconstructed later from memory or from a spreadsheet somebody maintains by hand.

02 · Setting up Claude Fable 5 in Claude Code and Claude Desktop

Claude models on Quanta work only from the official Claude Code and Claude Desktop clients, and Claude Code 2.1.285 or newer is required for Claude Code. That pairing is deliberate. Claude Fable 5 is used in the clients Anthropic builds for it, and Quanta routes those requests through its Anthropic native protocol.

The setup itself is short:

• › Create a Quanta account and use the $0.50 trial credit to place a first call.
• › Add prepaid balance using one of the published top-up amounts: $1.39, $6.94, $13.89, $27.78, $69.44, or $138.89.
• › Create a live key in the Quanta dashboard and point Claude Code or Claude Desktop at the Quanta API base address.
• › Select Claude Fable 5 (claude-fable-5) as the model and run the first request.

One key and one balance work across all three Quanta protocols: OpenAI-compatible, Anthropic native, and Gemini native. Gemini models use the native protocol through Gemini CLI. Claude traffic stays in the Claude clients, while the same Quanta account, key set, and balance cover the rest of your routing needs. Adding a second model family on Quanta does not mean opening a second billing relationship or learning a second console.

For teams, that matters at volume. A single dashboard shows the balance, the keys, the calls, and the charge attached to each call, so onboarding a new service or a new engineer does not require a separate procurement conversation.

03 · Rate limits and concurrency for sustained workloads

Heavy workloads rarely break because of a vague limit. Quanta publishes the limits it enforces, which lets a team plan capacity before a backlog arrives:

• › 120 requests per minute per key
• › 8 requests in flight per account
• › 25 MB per request body
• › 20 live keys per account
• › Streams have no total-time cap and end only after 300 seconds of upstream silence

These figures translate into concrete engineering decisions. Eight requests in flight per account means a batch job should be queued and paced rather than fired all at once; a team can spread work across up to 20 live keys per account, each carrying its own 120 requests per minute. A 25 MB request body gives a clear ceiling for how much context or payload a single call can carry, so oversized requests fail predictably instead of degrading quietly.

The stream rule is the one worth reading twice. Quanta does not impose a total-time cap on streams. A long generation is not cut off simply because it has been running for a while; the stream ends only after 300 seconds of upstream silence. For sustained, long-horizon jobs, that is a meaningful difference between a documented ceiling and an open-ended wait.

Quanta is designed around this kind of clarity. Concurrency, rate limits, and the token math are all stated, so the team planning the workload and the team reading the invoice are working from the same numbers.

04 · Prepaid balance, invoices, and per-call billing records

Quanta bills against a prepaid balance. There is no monthly fee and no minimum spend, so a team can fund exactly what it expects to use and stop there. Unused balance can be refunded within 15 days of the top-up, which keeps the prepaid model from becoming a lock-in.

For companies, payment runs by bank transfer from a company account. Invoices and advance quotes are arranged through the Contact page or the Messages page of the dashboard, so procurement can request the paperwork it needs in the same place the usage data already lives.

Two records support the financial side of the relationship. First, the dashboard lists the tokens and the charge of every call, which gives an audit trail down to the individual request. Second, the text you send is kept for 30 days for abuse handling and billing disputes, then deleted; model answers are not stored. That retention window is defined and published rather than left ambiguous.

Together, these details answer the questions a finance lead tends to ask about a relay service: what was billed, on what basis, where the evidence sits, and who can help with a quote or an invoice. Quanta answers them with dashboard records and two dashboard pages rather than a maze of channels.

05 · Monthly volume examples on Quanta

To make the shape of a Quanta bill concrete, here are examples calculated at a 3:1 input-to-output ratio with no cache hits. These are examples of what volume costs on Quanta, not quotes or estimates of any specific account.

• › Claude Opus 5.5: 10 million tokens a month ≈ $34.58; 100 million ≈ $345.83.
• › Claude Sonnet 5.5: 10 million tokens a month ≈ $25.07; 100 million ≈ $250.69.
• › GPT-6.1 Sol: 10 million tokens a month ≈ $4.10; 100 million ≈ $40.97.
• › Gemini 2.5 Flash: 10 million tokens a month ≈ $6.49; 100 million ≈ $64.93.

Claude Fable 5 sits at $4.35 input, $21.74 output, $5.35 cache write, and $0.3333 cache read per 1M tokens, so a team can place it against those examples on the same basis. For teams weighing neighbouring options, Quanta also lists Claude Fable 5.1 at $4.38, $21.88, $5.39, and $0.3333, and Claude Opus 5 at $1.97, $9.83, $2.42, and $0.1222. Every figure sits in one place, with the same four token classes, so a model change is a comparison rather than an investigation.

Because Quanta charges per token with no monthly fee and no minimum spend, the practical sequence for a new team is simple: start small, watch the per-call records, and scale the balance as the workload proves out. There is no commitment to justify before the first request.

06 · Funding a Quanta account, referrals, and refunds

New accounts receive $0.50 in trial credit on sign-up, enough to confirm that the key, the client pairing, and the model selection all work before anyone commits budget. Top-ups come in six published amounts, from $1.39 up to $138.89, and each one lands in the same prepaid balance that every call draws from.

Quanta also returns a share of the network it builds. Referral rewards are 19% of a referred account's top-ups, with 5% extending one level further. Rewards can be withdrawn once they reach $27.78, or moved into the balance to cover usage. That choice matters for teams that would rather spend the credit on traffic than cash it out.

Refunds remain straightforward: unused balance can be refunded within 15 days of the top-up.

For a team choosing where to run Claude Fable 5, the argument for Quanta is not a single number. It is the combination of published per-token prices, enforced limits that are stated in advance, per-call records in the dashboard, a prepaid balance with no monthly fee, and invoices arranged through the Contact page or the Messages page. Open a Quanta account, start with the $0.50 trial credit, and put Claude Fable 5 to work in Claude Code or Claude Desktop. If the workload grows, the billing already explains itself.