Hosted model directory
AI API pricing calculator
At 1.0M input and 0.2M billed output tokens per month, listed current rates range from $1.50 to $20.00. Compare dated vendor rates for hosted models, then decide whether an API or a local model fits your work.
Usage estimate
Estimate monthly token spend
Enter tokens in millions. The directory stays fully readable without JavaScript.
Showing 6 hosted models.
Estimates use uncached standard input and billed output rates. They exclude cache writes, tools, subscriptions, discounts, taxes and request-length surcharges. Expired rates are unavailable.
Vendor-sourced rates
Hosted models and listed API rates
Scroll sideways to view all rate columns.
| Model | API access | Input rate | Output rate | Monthly estimate |
|---|---|---|---|---|
| Claude Opus 5Anthropic · verified 2026-09-27 | claude-opus-5Active legacy model on the Claude API, with retirement no sooner than July 24, 2027. | $5.00 / M | $25.00 / M | $10.00 |
| Claude Opus 5.5Anthropic · verified 2026-09-27 | claude-opus-5-5Current Claude Opus model on the Claude API. | $4.00 / M | $20.00 / M | $8.00 |
| Claude Sonnet 5Anthropic · verified 2026-09-27 | claude-sonnet-5Current Claude API model. | $2.00 / M | $10.00 / M | $4.00 |
| Gemini 3.8 FlashGoogle · verified 2026-09-27 | gemini-3.8-flashGenerally available through the Gemini API and ready for production use. | $0.75 / M | $3.75 / M | $1.50 |
| GPT-6 AstraOpenAI · verified 2026-09-27 | gpt-6-astraAvailable through the OpenAI API. | $10.00 / M | $50.00 / M | $20.00 |
| Grok 4.7xAI · verified 2026-09-27 | grok-4.7Available through the xAI API. | $2.00 / M | $6.00 / M | $3.20 |
Rates are USD per million text tokens. Each rate links to its vendor source and carries a verification date. Expired rates are withheld from estimates.
Local or hosted
An API listing is not a local download
These entries describe vendor-hosted API access. They do not imply downloadable weights, local hardware fit, benchmark parity or a global ranking. For an open-weight alternative, check your machine in the local model catalog or start with local versus cloud.
Price scope and dates
Claude Opus 5: verified 2026-09-27. Standard Claude API rates. Cache writes, batch discounts, fast mode, and US inference pricing are excluded.
Claude Opus 5.5: verified 2026-09-27. Standard Claude API rates. Cache writes, batch discounts, fast mode, and US inference pricing are excluded.
Claude Sonnet 5: verified 2026-09-27. Standard Claude API rates. Cache writes, batch discounts, fast mode, and US inference pricing are excluded.
Gemini 3.8 Flash: verified 2026-09-27. Standard paid Gemini Developer API rates through December 31, 2026. Output includes thinking tokens; cache storage, tools, and other service tiers are excluded.
GPT-6 Astra: verified 2026-09-27. Standard text rates. Requests over 272K input tokens have higher rates; cache writes and tools are excluded.
Grok 4.7: verified 2026-09-27. Global short-context rates below 200K prompt tokens. Longer prompts, US regional inference, and tools cost more.
FAQ
Frequently asked questions
Is this an API bill quote?
No. Each estimate multiplies entered monthly input and billed output tokens by the linked vendor rate. It excludes subscriptions, tools, cache writes, discounts, taxes and request-length surcharges.
Can I run these hosted models locally?
No. This directory marks models that are offered through a provider API. Use the local catalog to check open-weight models against your hardware.
Why is a price unavailable?
A model has no listed standard token rate, or its listed rate has expired. Follow the provider price link for the current service terms.
Looking for the closed-weight boundary? Read Can you run Claude locally?. Download the dated model and price data.