Short answer: I could not confirm a current paid rate for the direct MiniMax-M2 API from the official documentation reviewed. MiniMax’s M2 repository says the API was “free for a limited time,” but gives no promotion dates, eligibility, limits, or later paid rate. That statement does not confirm the offer is still available. (MiniMax-M2 repository)
A third-party article quotes $0.15 per million input tokens and $0.60 per million output tokens through OpenRouter, dated October 2025. Treat these as historical, OpenRouter-specific figures—not a verified current rate for MiniMax’s direct API. (Apidog article)
Which MiniMax M2 price applies?
| Route or model | What the available evidence says | What it means for your budget |
|---|---|---|
| MiniMax-M2 direct API | The official M2 repository describes a limited-time-free offer but does not state a numeric rate or its terms. The official pricing and model-documentation excerpts reviewed here name M3 models; those excerpts do not establish whether M2 is unavailable or covered by separate or legacy terms. (MiniMax-M2 repository; pricing overview; model invocation guide) | A current direct M2 rate and an active free offer remain unconfirmed. Check the exact model and current terms with MiniMax before budgeting. |
| MiniMax M2 through OpenRouter | A third-party article reports $0.15/M input and $0.60/M output, as of October 2025. (Apidog article) | Historical intermediary pricing only; do not assume it remains available or applies to MiniMax’s own API. |
| MiniMax-M2.1 | A third-party listing reports $0.30/M input and $1.20/M output for M2.1. (LLM Stats) | M2.1 is a different model version; these figures do not establish M2 pricing. |
Here, M means one million tokens. Input and output tokens have separate rates, so a request count alone—or a blended price without a stated input/output mix—can mislead.
How to estimate a budget from token use
Use the applicable provider rates and estimated token volumes:
Estimated cost = (input tokens ÷ 1,000,000 × input rate) + (output tokens ÷ 1,000,000 × output rate).
For scale only, applying the historical OpenRouter quote of $0.15/M input and $0.60/M output to a hypothetical workload of 10 million input tokens and 2 million output tokens gives:
- Input: 10 × $0.15 = $1.50
- Output: 2 × $0.60 = $1.20
- Total: $2.70
This is an arithmetic example using the article’s October 2025 figures—not a current M2 quote, a MiniMax-direct API estimate, or observed usage. The same article estimates 5,000–7,000 requests per $1 of OpenRouter credit, but says the number depends on output length and supplies no token assumptions. That estimate is not a dependable budget for a specific workload. (Apidog article)
What to confirm before using M2 in production
- Confirm current availability and price for the exact route. The official invocation-guide excerpt reviewed here names M3 models, but does not establish that the excerpt is a complete list or that M2 is unavailable. Ask MiniMax whether the exact M2 API model remains available and whether any legacy terms apply. (Model invocation guide)
- Check the provider and exact model identifier. An intermediary’s M2 price is not automatically MiniMax’s direct API price. Make sure the listing is M2—not M2.1 or another version—and check that provider’s current billing terms.
- Verify any free offer rather than relying on old wording. The repository does not specify an end date, eligibility rules, or usage limits for its limited-time-free statement. Confirm current availability and terms before relying on it. (MiniMax-M2 repository)
- Check applicable capacity limits. MiniMax says API rate limits depend on the model and interface. The rate-limit documentation excerpt reviewed here does not establish which limits, if any, apply to M2; confirm the terms for the exact model and account. (Rate limits guide)
MiniMax describes M2 as designed for coding and agentic workflows and says its model weights are available for local deployment. Those descriptions do not establish current API availability, service terms, or suitability for a particular workload. Local deployment is a separate cost path; the repository does not quantify hardware or operating costs. (MiniMax-M2 repository)
Bottom line
The evidence reviewed here is not enough to set a current production budget for the direct MiniMax-M2 API: a current paid rate and the terms of any active free offer remain unconfirmed. The October 2025 OpenRouter figures are historical third-party claims, and M2.1 pricing must not be substituted for M2. Before committing to a budget, confirm the exact M2 listing, provider rate and billing unit, any free allowance, and the limits that apply to your account.
Sources
How To Use MiniMax M2 For Free
MiniMax M2.1 Benchmarks, Pricing & Context Window
MiniMax API Docs: Pricing Overview