In this article

Key takeaways
Documented- Answer: From August 25, 2026, Make calculates AI Provider credits for Small, Medium and Large models with separate input and output token rates. Check your steps.
- Evidence: Based on 1 dated primary or official source, most recently checked .
- Scope: This article does not claim hands-on testing. Performance or safety verdicts require a linked test record.
What changed
Make changed how it calculates credits for Small, Medium and Large models in Make's AI Provider. The change took effect on August 25, 2026. It affects users who run AI steps through Make's AI Provider. It does not affect apps where you use your own API key.
Event date
The Make Help Center article says "As of August 25, 2026" and is stamped "Updated 25 Aug 2026".
Details as the source states them
Make says it previously calculated credits for Small, Medium and Large models from total tokens, with input and output combined. Now input and output tokens have separate rates. Make states these rates, in tokens per credit (the previous rate for total tokens is shown in brackets):
- Small: 18,080 input tokens and 2,260 output tokens (previously 5,000 total tokens).
- Medium: 18,080 input tokens and 2,260 output tokens (previously 3,500 total tokens).
- Large: 3,616 input tokens and 452 output tokens (previously 1,500 total tokens).
Make says input tokens are significantly cheaper and output tokens are moderately more expensive. Its example: a data extraction step with 2,000 input tokens and 100 output tokens cost about 0.6 credits on the Medium tier under the previous rates. Under the new rates, the same step costs about 0.16 credits, which Make calls a 73% reduction.
Make says most use cases, such as extraction, summarization, classification and agent workflows, see lower consumption. Steps that generate long outputs from short prompts may use more credits. Make says no action is required: you can keep using Make's AI Provider or bring your own API key.
What this does not say
- The article does not give a price per credit. Credit prices depend on your plan.
- The 73% figure is Make's single worked example. Your own savings depend on your prompts and outputs.
What to check
Estimate the input and output tokens of your largest AI steps. Apply the rates above to see whether credit use goes up or down. Check Make's credits documentation for how credits count against your plan.
Affected profile
See the Anavem profile for Make. For background on selecting a platform, read how to choose an AI automation tool.
Source: Make Help Center, "More value for your credits: Updated Make's AI Provider token pricing model" (accessed 2026-10-03). Nothing here was tested by Anavem.
Tools mentioned
Sources
Related guides and updates
- NewsRelevance AI releases a new Invent builder assistantOct 3, 2026 · 2 min read
- NewsGumloop moves to pass-through pricing plus an 8% feeOct 3, 2026 · 2 min read
- NewsZapier deprecates Zapier Functions, shutdown September 1Oct 3, 2026 · 2 min read
- NewsZapier prices AI by Zapier steps by model tierOct 3, 2026 · 2 min read
- NewsRead AI becomes a Claude Connector (May 27, 2026)Oct 3, 2026 · 2 min read