LLM API cost calculator
Enter your monthly tokens and the input, cached and output mix, and compare what each model would cost.
Free, runs in your browser, no sign-up. Updated 2026-10-25.
Result
1,000M tokens a month at 37/58/5 input/cached/output: cheapest Claude Haiku 5.5 at $67.80, dearest gpt-6-astra at $6,780 (100x spread).
- Dearest per month
- $6,780
- Spread
- 100×
| Model | Monthly cost | Rates | Source |
|---|---|---|---|
| Claude Haiku 5.5Anthropic. Prompts up to 100,000 tokens. Longer prompts cost 0.50 / 0.05 / 2.50. | $67.80 | $0.1 / $0.01 / $0.5 | Checked 2026-10-23 |
| gpt-6-lunaOpenAI | $67.80 | $0.1 / $0.01 / $0.5 | Checked 2026-10-23 |
| MiniMax M3.1-FlashMiniMax | $94.20 | $0.15 / $0.015 / $0.6 | Checked 2026-10-23 |
| gpt-5.6-lunaOpenAI | $146 | $0.2 / $0.02 / $1.2 | Checked 2026-10-23 |
| deepseek-flashDeepSeek. Peak-hour price. Off-peak is half. | $174 | $0.3 / $0.006 / $1.2 | Checked 2026-10-23 |
| MiniMax M3MiniMax. Shown after a 'permanent 50% off' (list 0.60 / 0.12 / 2.40). Page layout was unclear: check again before launch. | $206 | $0.3 / $0.06 / $1.2 | Checked 2026-10-23 |
| Gemini 2.5 FlashGoogle | $253 | $0.3 / $0.03 / $2.5 | Checked 2026-10-23 |
| Gemini 3.8 FlashGoogle. Launch price until 31 Dec 2026, then double. | $508 | $0.75 / $0.075 / $3.75 | Checked 2026-10-23 |
| gpt-5.4-miniOpenAI | $546 | $0.75 / $0.075 / $4.5 | Checked 2026-10-23 |
| Claude Haiku 4.5Anthropic | $678 | $1 / $0.1 / $5 | Checked 2026-10-23 |
| deepseek-v4-proDeepSeek. Peak-hour price. Off-peak is half. | $712 | $1.32 / $0.044 / $3.96 | Checked 2026-10-23 |
| Gemini 2.5 ProGoogle | $1,035 | $1.25 / $0.125 / $10 | Checked 2026-10-23 |
| Claude Sonnet 5.5Anthropic | $1,298 | $2 / $0.1 / $10 | Checked 2026-10-23 |
| gpt-6.1-solOpenAI | $1,298 | $2 / $0.1 / $10 | Checked 2026-10-23 |
| Gemini 4 ArgonGoogle. Launch price until 30 Nov 2026. | $1,298 | $2 / $0.1 / $10 | Checked 2026-10-23 |
| Claude Sonnet 5Anthropic | $1,356 | $2 / $0.2 / $10 | Checked 2026-10-23 |
| gpt-5.6-terraOpenAI | $1,456 | $2 / $0.2 / $12 | Checked 2026-10-23 |
| Gemini 3.1 Pro PreviewGoogle | $1,456 | $2 / $0.2 / $12 | Checked 2026-10-23 |
| Claude Opus 5.5Anthropic | $2,596 | $4 / $0.2 / $20 | Checked 2026-10-23 |
| gpt-5.6-solOpenAI | $2,712 | $4 / $0.4 / $20 | Checked 2026-10-23 |
| Claude Opus 5Anthropic | $3,390 | $5 / $0.5 / $25 | Checked 2026-10-23 |
| Claude Fable 5.1Anthropic | $6,345 | $10 / $0.25 / $50 | Checked 2026-10-23 |
| gpt-6-astraOpenAI | $6,780 | $10 / $1 / $50 | Checked 2026-10-23 |
Assumptions
Every number the result depends on, and where it comes from. Change any value and the result updates.
| Assumption | Value | Source |
|---|---|---|
| Uncached input share of tokens | % of tokens | OceanLab assumption. Example mix measured on one production workload of about 48,000 calls. Use your own.Reviewed 2026-10-25 |
| Cached input share of tokens | % of tokens | OceanLab assumption. Same example workload. Cache reads are billed at the cached rate; cache writes are not modelled.Reviewed 2026-10-25 |
| Output share of tokens | % of tokens | OceanLab assumption. Same example workload.Reviewed 2026-10-25 |
| Model prices (23 models) | 23 | Each provider's pricing page, linked per row belowChecked 2026-10-23 |
Send this comparison to OceanLab
Opens an email to OceanLab with this result and a link to your inputs. Nothing is sent until you press send, and there is no sign-up.
How this is calculated
Cost = tokens × (input share × input price + cached share × cached input price + output share × output price) ÷ 1,000,000.
Prices are standard pay-as-you-go rates for short context, read from each provider's own pricing page on the date shown. Cache write charges and batch discounts are not included.
The default mix of 37% input, 58% cached input and 5% output is an example from one production workload. Use your own split: the cheapest model can change with the mix alone.
Sources
- Anthropic API pricing · checked 2026-10-23
- OpenAI API pricing · checked 2026-10-23
- Google API pricing · checked 2026-10-23
- DeepSeek API pricing · checked 2026-10-23
- MiniMax API pricing · checked 2026-10-23
- ECB euro reference exchange rates (22 Oct 2026) · checked 2026-10-23
Questions about this tool
Why does the token mix matter so much?
Input, cached input and output are priced differently. A model with cheap output but expensive cache reads can win or lose on the mix alone, so the ranking can flip.
Where do the prices come from?
Each row links to the provider's official pricing page and shows the date it was checked. They are standard, short-context rates.
Are batch and off-peak discounts included?
Off-peak pricing is included for providers that publish it, when you switch it on. Batch discounts and cache write charges are not modelled.
Is anything I enter sent anywhere?
No. The tool runs in your browser and there is no sign-up. Nothing leaves the page unless you choose to email your result to OceanLab.
Other free tools
OceanLab product builds like this
Products OceanLab designed and built. Each case study covers the screens, the scope and the rules it was built for.
Sonde - LLM Evaluation and Observability Platform with Release Gates
Designed and built a clickable LLM evaluation and observability platform for the AI team of a French health mutual: redacted production traces span by span, release gates that block a prompt or model change when quality regresses, alerts with owners and ageing, and the monitoring record its risk and compliance teams can show a supervisor.
Kaganek - RAG Knowledge Assistant with Citations and Permissions Intact
Designed and built a clickable internal knowledge assistant for a Kraków consulting engineering firm: staff ask in Polish or English and get answers built only from the firm's own SharePoint, Confluence, Google Drive and file share documents, with every claim linked to its source passage and an honest "I don't know" when the sources are weak.
Tekmė - No-Code Workflow Automation Builder with AI Steps
Designed and built a clickable no-code automation builder for a Vilnius SaaS startup selling to EU operations teams: a node canvas with branches, error paths and AI steps that classify, extract, summarise or act as an agent, a cost estimate per run, and a run history where every step is recorded with personal data masked.
Lektor - AI Code Review Agent That Comments but Never Merges, Hosted Inside a Private Bank
Designed and built a clickable AI code review agent for the engineering team of a Zurich private bank: it reviews every merge request on GitLab and GitHub Enterprise with models hosted in the bank's own tenancy, masks client data and secrets before any model sees the diff, posts inline findings with rule, CWE and suggested patch, and never approves or merges.
