OpenAI
GPT-6 Luna
$10.00/ month
$0.001 per request
- Input
- $5.00
- Cache reads / writes
- $0.00 / $0.00
- Output
- $5.00
FREE TOOL · BY APIDECK
LLM cost calculator.
Know your API bill.
LLM cost calculator for OpenAI and Claude. Enter your workload once and compare monthly token costs, with every assumption in view.
ILLUSTRATIVE EXAMPLE
Monthly API cost
10,000 requests × 5,000 input + 1,000 output tokens per request. Cache reads 0%; writes 0%.
Anthropic
Claude Haiku 5.5
$10.00/ month
$0.001 per request
- Input
- $5.00
- Cache reads / writes
- $0.00 / $0.00
- Output
- $5.00
OpenAI
GPT-6.1 Terra
$100.00/ month
$0.01 per request
- Input
- $50.00
- Cache reads / writes
- $0.00 / $0.00
- Output
- $50.00
OpenAI
GPT-6.1 Sol
$200.00/ month
$0.02 per request
- Input
- $100.00
- Cache reads / writes
- $0.00 / $0.00
- Output
- $100.00
Anthropic
Claude Sonnet 5.5
$200.00/ month
$0.02 per request
- Input
- $100.00
- Cache reads / writes
- $0.00 / $0.00
- Output
- $100.00
Anthropic
Claude Opus 5.5
$400.00/ month
$0.04 per request
- Input
- $200.00
- Cache reads / writes
- $0.00 / $0.00
- Output
- $200.00
Sorted by estimated cost, not capability. Text API usage only; tools, taxes and other charges are excluded. Estimates use the same token counts across models, although tokenizers differ.
OTHER WAYS TO BUY
Could a router lower your bill?
LLM routers may offer promotional credits, alternative provider routes or negotiated rates. Compare the final bill, including platform fees, minimum spend and the price after a promotion ends.
Promotional credits
Check eligible models, new-customer restrictions and expiry dates.
Alternative rates
Match the exact model, provider route and service tier before comparing.
Total cost
Include credit-purchase fees, commitments and post-promotion pricing.
Building a model or router? Send us pricing corrections, new model listings, promotional credits or marketing partnership proposals. Pricing corrections do not require a commercial partnership.
Compare LLM routers →Submit an update or partnership ↗
No partner offers are applied to the estimates above. Submit a public offer with eligibility and expiry terms for review. Sponsored or affiliate placements will be labeled and will not change the direct-provider cost ranking. Sponsorship policy →
Explore models on OpenRouter
16 publisher-corroborated model listings. Search across model providers, with available models from OpenRouter’s top ten usage ranking first. Router prices are separate from the direct-provider comparison above.
Source: OpenRouter rankings, usage through October 25, 2026. Licensed under CC BY 4.0. Popularity reflects OpenRouter token traffic, not quality or global market share. Variants absent from the pricing catalog are not substituted.
16 matching models · showing 12
DeepSeek: DeepSeek V4.1 Flash
$4.293/ month
Token-only estimate using the submitted workload above. Free variants may have rate limits.
- Input / 1M tokens
- $0.0477
- Output / 1M tokens
- $0.1908
Explore provider API profile →
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Z.ai: GLM 5.3 Flash
$12.50/ month
Token-only estimate using the submitted workload above. Free variants may have rate limits.
- Input / 1M tokens
- $0.15
- Output / 1M tokens
- $0.50
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Tencent: Hy4 preview
Check route pricing
An estimate is unavailable for this workload: check cache rates, token limits, additional charges or variable pricing at the source.
- Input / 1M tokens
- $0.7506
- Output / 1M tokens
- $2.2509
Pricing varies by tier or time.
Check model and route details ↗Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
OpenAI: GPT-6 Luna
Check route pricing
An estimate is unavailable for this workload: check cache rates, token limits, additional charges or variable pricing at the source.
- Input / 1M tokens
- $0.10
- Output / 1M tokens
- $0.50
Pricing varies by tier or time.
Check model and route details ↗Explore provider API profile →
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Anthropic: Claude Opus 5.5
Check route pricing
An estimate is unavailable for this workload: check cache rates, token limits, additional charges or variable pricing at the source.
- Input / 1M tokens
- $4.00
- Output / 1M tokens
- $20.00
Explore provider API profile →
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
NVIDIA: Nemotron 3 Ultra (free)
$0.00/ month
Token-only estimate using the submitted workload above. Free variants may have rate limits.
- Input / 1M tokens
- $0.00
- Output / 1M tokens
- $0.00
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Z.ai: GLM 5.3
$17.495/ month
Token-only estimate using the submitted workload above. Free variants may have rate limits.
- Input / 1M tokens
- $0.0699
- Output / 1M tokens
- $1.40
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Anthropic: Claude Haiku 5.5
Check route pricing
An estimate is unavailable for this workload: check cache rates, token limits, additional charges or variable pricing at the source.
- Input / 1M tokens
- $0.10
- Output / 1M tokens
- $0.50
Pricing varies by tier or time.
Check model and route details ↗Explore provider API profile →
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Anthropic: Claude Sonnet 5.5
Check route pricing
An estimate is unavailable for this workload: check cache rates, token limits, additional charges or variable pricing at the source.
- Input / 1M tokens
- $2.00
- Output / 1M tokens
- $10.00
Explore provider API profile →
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Google: Gemini 4 Argon
Check route pricing
An estimate is unavailable for this workload: check cache rates, token limits, additional charges or variable pricing at the source.
- Input / 1M tokens
- $2.00
- Output / 1M tokens
- $10.00
Explore provider API profile →
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
MiniMax: MiniMax M3.1 Flash
$13.50/ month
Token-only estimate using the submitted workload above. Free variants may have rate limits.
- Input / 1M tokens
- $0.15
- Output / 1M tokens
- $0.60
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
Mistral: Mistral Large 4
$109.80/ month
Token-only estimate using the submitted workload above. Free variants may have rate limits.
- Input / 1M tokens
- $1.36
- Output / 1M tokens
- $4.18
Explore provider API profile →
Publisher model evidence ↗ · Reviewed 2026-10-26. Model existence is corroborated; router availability and prices are reported by OpenRouter, not runtime-tested.
452 imported listings are withheld pending publisher review; this does not mean they are fake. Pricing snapshot: 2026-10-26, from the OpenRouter Models API. Prices can vary by route and change over time. Fees, tools and taxes are excluded. Models with variable rates or additional declared charges do not receive a calculated total. Daily automated checks propose reviewed snapshot updates; this is not a live feed.
Direct-provider pricing you can check
USD per million tokens. Standard direct-provider processing, within the calculator’s 100,000-token combined request scope. No batch or regional adjustments. Daily automated source checks propose updates for review before publication. This is a dated snapshot, not a live feed.
| Model | Input | Output | Cache read | Cache write | Publisher source |
|---|---|---|---|---|---|
| GPT-6.1 Sol | $2.00 | $10.00 | $0.10 | $2.5030 minutes | Check pricing ↗ |
| GPT-6.1 Terra | $1.00 | $5.00 | $0.05 | $1.2530 minutes | Check pricing ↗ |
| GPT-6 Luna | $0.10 | $0.50 | $0.01 | $0.12530 minutes | Check pricing ↗ |
| Claude Opus 5.5 | $4.00 | $20.00 | $0.20 | $5.005 minutes | Check pricing ↗ |
| Claude Sonnet 5.5 | $2.00 | $10.00 | $0.10 | $2.505 minutes | Check pricing ↗ |
| Claude Haiku 5.5 | $0.10 | $0.50 | $0.01 | $0.1255 minutes | Check pricing ↗ |
Prices are a dated snapshot. Recheck the publisher source before committing a budget. Cache writes are included only when you allocate a share of input to them.
From estimate to your first request
API-key setup guides from the Apideck team.
How the estimate works
Monthly cost = requests × (uncached input tokens × input rate + cached tokens × cache-read rate + cache-write tokens × write rate + output tokens × output rate) ÷ 1,000,000.
How do you calculate LLM API costs?
Multiply monthly requests by the input and output tokens per request and their respective per-million-token rates. Split input tokens into uncached input, cache reads and cache writes when applicable, then add the four costs.
Does this include ChatGPT or Claude subscriptions?
No. These are standard direct-provider text API estimates in USD. Chat subscriptions, taxes, tools, web searches, storage, images, audio, regional uplifts and negotiated discounts are excluded. Batch, fast and long-context pricing are not modeled.
Are the cheapest models equally capable?
No. This compares the cost of the same token workload, not quality, latency or success rate. Tokenizers differ across models, so use actual provider usage measurements for a more accurate comparison. Include conversation history, tool schemas and billable reasoning tokens in your estimates.
How is prompt caching priced?
Cache reads replace part of the ordinary input charge. Cache writes use their own rate. This calculator uses 30-minute writes for the listed OpenAI models and 5-minute writes for Claude. Cache eligibility and hit rates depend on the provider; no cache savings are assumed by default.
How current are these prices?
This is a dated snapshot, not a live price feed. Daily automated checks propose updates for review. Check the linked publisher pricing before budgeting. The calculator covers requests with at most 100,000 combined input and output tokens; provider token and output limits still apply.