Free public price board
LLM list prices, updated monthly.
Look up a published rate for OpenAI, Claude, Gemini, Grok, or DeepSeek. Download the tool to quote a token count on your machine, or to mask an email, a phone number, or an IPv4 address before you send the text.
Prices are vendor list prices in USD, refreshed on the 1st of each month. A cheaper row is a cheaper rate. There is nothing to buy.
Look up a rate
Short-context list prices for the models we track, with the vendor page each number came from.
Quote a request
Give an input and output token count. The tool prices every model in the catalog. It does not call the model.
Redact before you send
Regex masks for emails, phone numbers, and IPv4 addresses. Review the text yourself. Names are not detected.
On your machine
llm-cost-opt quote --input-tokens 1500 --output-tokens 300
llm-cost-opt anonymize --text "Email me at me@example.com"The rates
US dollars per 1,000,000 tokens, short context. The date under the table is the last refresh. A cheaper row is a cheaper published rate.
| Model ⇅ | Provider ⇅ | Input ($/1M) ↑ | Output ($/1M) ⇅ | What it's for |
|---|---|---|---|---|
| gpt-6-luna Cached input $0.01 / 1M | OpenAI | $0.10 | $0.50 | High-volume cheap OpenAI |
| claude-haiku-5.5 Short-context rate for prompts up to 100k tokens | Anthropic | $0.10 | $0.50 | Fast Anthropic workhorse |
| deepseek-flash Peak cache-miss; off-peak is half | DeepSeek | $0.30 | $1.20 | Lowest DeepSeek cost |
| gemini-3.8-flash Promo through December 31, 2026; then $1.50 / $7.50 | $0.75 | $3.75 | Cheapest capable Flash | |
| grok-4.3 | xAI | $1.25 | $2.50 | Cheaper Grok tier |
| deepseek-v4-pro Peak cache-miss; off-peak is half | DeepSeek | $1.32 | $3.96 | Strong value reasoning |
| gpt-6.1-sol Cached input $0.10 / 1M | OpenAI | $2.00 | $10.00 | Daily driver coding & chat |
| claude-sonnet-5.5 | Anthropic | $2.00 | $10.00 | Coding & agents balance |
| gemini-3.1-pro <=200k prompt tokens | $2.00 | $12.00 | Multimodal / long context | |
| grok-4.7 | xAI | $2.00 | $6.00 | xAI flagship general |
| claude-opus-5.5 | Anthropic | $4.00 | $20.00 | Hard agentic coding |
| gpt-6-astra Cached input $1 / 1M | OpenAI | $10.00 | $50.00 | Flagship reasoning / agents |
| claude-fable-5.1 | Anthropic | $10.00 | $50.00 | Long-running agents |
Last Updated: 2026-10-19 (standard short-context list prices). Refreshes on the 1st of each month.
Sources: OpenAI, Anthropic, Google Gemini, xAI, DeepSeek public pricing. Claude calls use the Fable, Opus, Sonnet, and Haiku families. Click a provider for its official page.
The last column is a short description we wrote. It is not a quality score.
A month of calls, at list price
Enter a volume and a token count. The calculator shows always using the flagship next to an example mix of cheaper models. Same tokens, different published rates.
Regex redaction
Runs in the browser demo and in the local CLI. Nothing is uploaded for the demo.
Emails
Pattern match
Phone numbers
Common formats
IPv4
Dotted quads
Long numbers
Optional, off by default
Before
Contact: john.doe@company.com Phone: +1-555-123-4567 Server: 192.168.1.100
After
Contact: [EMAIL_1] Phone: [PHONE_1] Server: [IP_1]
$ llm-cost-opt anonymize --text "Email me at me@example.com"
Regex will miss names, free-text addresses, and many identifiers. Read the output. This is not a compliance product.