SourceVane · Practical AI
AI API pricing comparison
Compare selected first-party text API list prices using the same per-million-token unit, with conditions, verification dates and official sources.
Verified selection
11 text API price records
USD per 1 million tokens. Sort mentally only within matching conditions; provider discounts, taxes and tools are excluded.
| Provider and model | Route / mode | Input | Cached input | Cache write | Output | Conditions | Checked |
|---|---|---|---|---|---|---|---|
OpenAIgpt-6-astra | first party api standard · short | $5.00 | $0.500 | $6.25 | $25.00 | Standard processing, short-context tier, global first-party API | Official source |
OpenAIgpt-5.6-sol | first party api standard · short | $2.00 | $0.200 | $2.50 | $10.00 | Standard processing, short-context tier, global first-party APIOpenAI describes this as promotional pricing available at least through the stated date. | Official source |
OpenAIgpt-5.6-luna | first party api standard · short | $0.100 | $0.010 | $0.125 | $0.600 | Standard processing, short-context tier, global first-party API | Official source |
Anthropicclaude-fable-5-1 | first party api standard · standard | $10.00 | $0.250 | $12.50 | $50.00 | Standard, first-party API, text tokensCache write shown is the 5-minute cache-write rate. | Official source |
Anthropicclaude-opus-5 | first party api standard · standard | $5.00 | $0.500 | $6.25 | $25.00 | Standard, first-party API, text tokensCache write shown is the 5-minute cache-write rate. | Official source |
Anthropicclaude-sonnet-5 | first party api standard · standard | $2.00 | $0.200 | $2.50 | $10.00 | Standard, first-party API, text tokensCache write shown is the 5-minute cache-write rate. | Official source |
Anthropicclaude-haiku-4-5 | first party api standard · standard | $1.00 | $0.100 | $1.25 | $5.00 | Standard, first-party API, text tokensCache write shown is the 5-minute cache-write rate. | Official source |
Googlegemini-3.8-flash | first party api standard · standard | $0.750 | $0.075 | — | $3.75 | Paid tier, standard processing, introductory rate through the stated dateThe output rate includes thinking tokens. Context-cache storage is billed separately. | Official source |
Googlegemini-3.5-flash-lite | first party api standard · standard | $0.300 | $0.030 | — | $2.50 | Paid tier, standard processing; input covers text, image, video and audioThe output rate includes thinking tokens. Context-cache storage is billed separately. | Official source |
DeepSeekdeepseek-flash | first party api standard · peak | $0.300 | $0.006 | — | $1.20 | Peak hours, cache-miss input, Monday–Friday 01:00–04:00 and 06:00–10:00 UTCThe displayed input price is the cache-miss rate. Cache-hit price is listed separately. | Official source |
DeepSeekdeepseek-v4-pro | first party api standard · peak | $1.32 | $0.044 | — | $3.96 | Peak hours, cache-miss input, Monday–Friday 01:00–04:00 and 06:00–10:00 UTCThe displayed input price is the cache-miss rate. Cache-hit price is listed separately. | Official source |
Use a price in a workload
The calculator can prefill standard text rates from this same catalogue. It keeps success rate and retries visible because price per token alone does not decide the useful cost.
Continue this decision
Evidence and next steps
Reduce AI API costs without losing useful results
Measure workload, retries and accepted outputs before comparing models, batching or reusable prompts.
Prompt-cache break-even: how many prefix reuses cover the write cost?
A source-backed calculation compares one cache write plus cache reads with ordinary input processing for a fixed 10,000-token prefix.
AI voice API pricing comparison
Compare selected voice-session and transcription prices while keeping unlike billing units and input/output components separate.
Measure AI API retry cost before changing models
Count failed attempts, final successful tasks and token use separately so retries do not hide the real cost of useful API work.
What happened next · API prices · Decision guides · Latest AI coverage · Follow updates