Cycles

Appearance

Claude vs GPT Cost Calculator ​

A free interactive calculator that compares per-call, per-day, per-month, and per-year cost across the major Claude and OpenAI models for any token volume and call rate.

Tip: Open fullscreen ↗ for a wider table, share/export buttons, and a shareable URL that preserves your configuration. The same toolbar is also available below.

USD · Standard API rates · GPT rates assume at most 272K input tokens per call. For longer prompts, enter the provider's long-context rates manually.

ModelInput $ / MOutput $ / MPer callPer dayPer monthPer year
$0.0600$600$18,000$219,000
$0.0120$120$3,600$43,800
$0.0006$6.00$180$2,190
$0.0120$120$3,600$43,800
$0.0600$600$18,000$219,000
$0.0240$240$7,200$87,600
$0.0120$120$3,600$43,800
$0.0060$60.00$1,800$21,900

Default rates last verified on 2026-09-30 against OpenAI and Anthropic; edit any cell to match your contracted rate. Shared links keep their saved rates, which may be older. Calculation: (input_tokens × input_$/M + output_tokens × output_$/M) ÷ 1,000,000 × calls/day. Excludes cache reads/writes, batch/flex discounts, fine-tuning, fast/ultrafast modes, tool fees, and data-residency premiums. Rates do not adjust automatically for long context: above 272K input tokens, the listed GPT models charge 2× input and 1.5× output rates. Claude Fable 5.1, Opus 5.5, and Sonnet 5.5 retain standard rates across their 1M context windows. Haiku 4.5 has a 200K context window; check model limits before budgeting a request. Include billed reasoning/thinking tokens in output totals. Token counts vary by tokenizer; the same text can use different token counts across models.

How the calculation works ​

Default model rates were verified on September 30, 2026 against OpenAI API pricing and Anthropic API pricing. The lineup includes GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna, GPT-6 Sol, Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5, and Claude Haiku 4.5.

Prices are in USD for standard API processing without caching. GPT rates assume at most 272,000 input tokens per call; each model's context limit still applies. Model prices change frequently; edit any rate to match current provider pricing or your contract. Shared URLs preserve saved models and rates, so an older link can contain older prices.

The cost for a single LLM call is:

cost_per_call = (input_tokens × input_price_per_M + output_tokens × output_price_per_M) ÷ 1,000,000

Per-day, per-month, and per-year columns multiply by calls_per_day, then by 30 and 365 respectively.

Use total billed output tokens, including OpenAI reasoning tokens or Claude thinking tokens, rather than just the visible answer. Token counts depend on the model's tokenizer, so identical text can produce different counts. The cheapest positive per-year total is highlighted; equal token volumes compare rates, not model quality or equal-text costs.

What the calculator does not include ​

  • Prompt caching. Reads have discounted rates, while cache writes can cost more than ordinary input. Savings depend on the model, cache lifetime, and hit rate; the calculator uses uncached input rates.
  • Batch and Flex discounts. Both providers publish 50% Batch discounts for the default models. OpenAI also offers Flex pricing. These modes are not selected automatically.
  • Fast and Ultrafast modes, data residency, and tool fees. These can add charges beyond standard token rates.
  • Fine-tuning costs. Per-token rates differ for fine-tuned model variants.
  • Reserved or committed-use pricing. Enterprise contracts often beat list pricing materially.
  • Automatic context-window pricing tiers. Above 272,000 input tokens, the default GPT models charge 2× the input rate and 1.5× the output rate for the entire request. Enter those rates manually for long-context workloads. Claude Fable 5.1, Opus 5.5, and Sonnet 5.5 have standard rates throughout their 1M-token context windows; Haiku 4.5 has a 200K-token context window. Check model limits before budgeting a request.

For accurate enterprise planning, treat the calculator as a directional estimate and verify with your provider account manager.

Why estimates are not the same as runtime authority ​

A common pattern: a team uses a calculator like this to project monthly cost at $4,000, sets up an alert for "$5,000 exceeded," and then loses $40,000 in a weekend to an agent that loops while the on-call team sleeps.

Calculators answer "what *might this workload cost?" Cycles can enforce caller-configured budgets at a mandatory application boundary before protected work begins. Application authorization separately decides which calls and tool arguments are allowed. Cost is one dimension of the resulting control design:

  • Blast radius. A single agent action — a deploy, an email blast, a database mutation — can cost more in damage than the agent's entire month of LLM bills. The host must authorize the tool and arguments; Cycles can meter caller-assigned action exposure alongside that decision.
  • Risk-weighted budgeting. Not every tool call is equal. RISK_POINTS lets the caller meter a team-defined exposure score separately from token cost; it is not an authorization decision.
  • Multi-tenant boundaries. A noisy tenant cannot drain shared headroom from quiet ones. See Multi-tenant SaaS.

If your projected $4,000/month is the actual constraint, the right response is not an alert at $5,000 but a pre-execution gate that will not let calls proceed beyond the cap — and an action-authority layer that prevents the catastrophic single mistake that no cost calculator can predict.