DeepSeek
Lowest55 in · 11 out
Planning estimate
- Input
- $0.000008
- Cached input
- $0.00
- Output
- $0.000003
- Daily
- $0.0108
- Monthly
- $0.3234
- $ / 1k tok
- $0.000163
- Budget fit
- 46,382,189
- Per request
- $0.000011
Calculator
Multi-model API cost calculator
Paste text to count tokens and compare providers side by side—or use the advanced volume planner for request-scale budgets.
- Characters
- 222
- Words
- 36
- Input tokens
- 54
- Output tokens
- 11
OpenAI / Grok / Perplexity cards use exact BPE (Planning estimate). Anthropic, Google, and DeepSeek use planning estimates — each card shows its own count.
Expected output size
20% of input when using presets
Volume & budget
Project spend and see how many requests fit your monthly budget.
Monthly requests30,000
Fits on lowest model46,382,189 req
Budget used0%
Planning insights
Actionable reads from this prompt + volume—not financial advice.
- Routing to DeepSeek V4 Flash instead of Claude Sonnet 4.5 saves $0.000322 per request ($9.67/mo at your volume).
- Try raising cached input share for RAG or system-prompt heavy traffic—cache rates can cut input spend when providers support it.
- Input-heavy workload: prefer models with cheaper input or higher cache discounts, and trim retrieved context.
- At 30,000 requests/mo, check batch pricing in the volume planner when latency allows.
- At $500.00/mo you can afford ~46,382,189 requests on DeepSeek V4 Flash (46,352,189 headroom vs current volume).
Special
Route & runway playbook
Split traffic between your cheapest and most expensive selected models, then see monthly savings and how many days your budget lasts—share the scenario without pasting your prompt.
Economy70% → DeepSeek V4 Flash
DeepSeek · $0.000011/req
Premium30% → Claude Sonnet 4.5
Anthropic · $0.000333/req
Blended / request$0.000107
Blended / month$3.22
Saved vs all-premium$6.77(68%)
Budget runway4,653 days
At 1,000 req/day on this mix
Cost by provider
53 in · 11 out
Planning estimate
- Input
- $0.000016
- Cached input
- $0.00
- Output
- $0.000027
- Daily
- $0.0434
- Monthly
- $1.30
- $ / 1k tok
- $0.000678
- Budget fit
- 11,520,737
- Per request
- $0.000043
$0.000033 more per request than the lowest option ($0.9786/mo).
Perplexity
+503%54 in · 11 out
Planning estimate
- Input
- $0.000054
- Cached input
- —
- Output
- $0.000011
- Daily
- $0.065
- Monthly
- $1.95
- $ / 1k tok
- $0.001
- Budget fit
- 7,692,307
- Per request
- $0.000065
$0.000054 more per request than the lowest option ($1.63/mo).
OpenAI
+1547%54 in · 11 out
Planning estimate
- Input
- $0.000068
- Cached input
- $0.00
- Output
- $0.00011
- Daily
- $0.1775
- Monthly
- $5.32
- $ / 1k tok
- $0.002731
- Budget fit
- 2,816,901
- Per request
- $0.000177
$0.000167 more per request than the lowest option ($5.00/mo).
xAI (Grok)
+2933%54 in · 11 out
Planning estimate
- Input
- $0.000162
- Cached input
- $0.00
- Output
- $0.000165
- Daily
- $0.327
- Monthly
- $9.81
- $ / 1k tok
- $0.005031
- Budget fit
- 1,529,051
- Per request
- $0.000327
$0.000316 more per request than the lowest option ($9.49/mo).
Anthropic
+2989%56 in · 11 out
Planning estimate
- Input
- $0.000168
- Cached input
- $0.00
- Cache write (5m)
- off
- Output
- $0.000165
- Daily
- $0.333
- Monthly
- $9.99
- $ / 1k tok
- $0.00497
- Budget fit
- 1,501,501
- Per request
- $0.000333
$0.000322 more per request than the lowest option ($9.67/mo).
Uncached input shown as 54 tokens at the OpenAI-style baseline; each card re-estimates with its family. Volume figures multiply the per-request total. Rates updated 2026-07-31. Approximate static list prices for planning only. Always verify against each provider’s official pricing page before production budgeting.
Advanced volume planner
Plan multi-request workloads with cached input, batch pricing, and model multi-select—same engine as before.
Usage inputs
Models to compare
Select up to 4
Cost estimate
1,000 requests · 2,000 in / 500 out
GPT-4o
OpenAI
$10.00
- Input
- $5.00
- Cached input
- $0.00
- Output
- $5.00
- Per request
- $0.01
Effective rates: in $2.50/1M · out $10.00/1M
Rates last updated 2026-07-31. Approximate static list prices for planning only. Always verify against each provider’s official pricing page before production budgeting.
Guide
How to get value from this calculator
Estimate monthly API spend across major LLM providers before you ship. Enter input and output token volumes separately, because output tokens often cost 3–5x more than input on many models. Use this as a planning baseline and verify against each provider's official pricing page.
Cost formula
Total cost = (input tokens / 1,000,000 × input price per 1M) + (output tokens / 1,000,000 × output price per 1M)
Why it matters
Most teams underestimate LLM bills because they track requests, not tokens, and ignore the output-side price premium. A realistic input/output split prevents surprise overruns when you scale from prototype traffic to production monthly volume.
How to use it
- Select the model or provider you plan to use in production.
- Enter your expected monthly input token volume from prompts, system messages, and retrieved context.
- Enter your expected monthly output token volume from completions and tool responses.
- Review the split between input and output cost to see where spend concentrates.
- Compare two or more models side by side for the same workload.
- Cross-check totals against the provider's current pricing page before budgeting.
Planning tips
- Model output tokens at 1.5–3x input volume for chat apps; code generation can run higher.
- Add 10–20% headroom for retries, eval runs, and staging environments.
- If your provider supports prompt caching, run the cache savings tool to adjust input assumptions.
- Batch API pricing can cut realtime cost by 50% on eligible workloads—compare both modes.
- Use the Route & runway playbook to model a cheap/premium traffic split and copy a shareable decision receipt.
- Re-run estimates when you change models; a cheaper input rate does not always mean lower total cost.
Frequently asked questions
Are these prices official?
No. CentsPerToken uses approximate static list prices for planning only. Always confirm current rates on each provider's pricing page before committing to a budget.
Why are input and output priced differently?
Providers charge separately because generating output tokens consumes more compute than processing input. On many frontier models, output per-million pricing is several times higher than input.
Should I include cached tokens in my input volume?
Cached prompt tokens are billed at a reduced rate on supported providers. Use the cache savings calculator to model hit rates instead of counting cached tokens at full input price.
How do I convert daily traffic to monthly tokens?
Multiply daily requests by average input and output tokens per request, then multiply by 30 for a rough monthly figure. Add peak-day buffer if traffic is spiky.