MiniMax API Pricing: M2.7 Text Tokens Versus the Separate Coding Plan

Pay-as-you-go model dimensions

The MiniMax first-party pricing reference lists M2.7 standard text inference at $0.30 per million input tokens and $1.20 per million output tokens, with $0.06 cache reads and $0.375 cache writes. Its M2.7-highspeed variant lists $0.60 input, $2.40 output, $0.06 cache reads and $0.375 writes. Rate rows should say both the variant and token class. A “half price” marketing interpretation would be incorrect if the user selects highspeed.

One coding plan can span several modalities

MiniMax’s coding Token Plan page describes Plus, Max and Ultra subscriptions with shared allowances across language models and other covered media features. For example its annual-billing presentation shows $200, $500 and $1,200 per year while its monthly comparison labels $20, $50 and $120 per month. These are different commitments, and estimates such as “calls per five hours” are not a contract for unlimited production API calls. Never convert this allowance into a public API dollar discount automatically.

Cache writes and reads are asymmetrical

Suppose an M2.7 job has 300,000 new input, 300,000 cache-read and 300,000 cache-write tokens plus 100,000 output tokens. The metered standard tariff arithmetic is 0.3×$0.30 + 0.3×$0.06 + 0.3×$0.375 + 0.1×$1.20 = $0.3405. Not every request has all four fields, and the price example must not be used if another model or highspeed service was actually called. Capture billing-side usage by category.

Long prompts and priority are not free

Other MiniMax models and service tiers can have their own context bands or traffic restrictions. The product page also differentiates audio and video models, which charge in characters, seconds or separate material inputs. Keeping everything under one “tokens” column will lead to poor budgeting. The API console, source version, model identifier, account region and priority setting should accompany the rate snapshot; private enterprise pricing remains unknown unless quoted.

What to include in a purchasing decision

For a coding agent, count platform subscription, external model calls, tool invocations, retries, concurrency and the number of editors. For a server application, use the public API meters and a budget cap rather than assuming that an interactive subscription covers a backend production system. Check whether a discount is long-lived or promotional: the vendor’s current display is not enough to reconstruct the exact day each model price began.

How to verify the actual invoice

Begin with a small representative workload rather than the cheapest published headline. Record the exact provider and billing channel, API model identifier, currency, account geography, plan and commitment length. The price or quota evidence in this guide is keyed to annual_subscription, cache_write_tokens_per_1m, cached_input_tokens_per_1m, input_tokens_per_1m, output_tokens_per_1m. Do not combine proprietary creative credits, token meters or media seconds into a single currency or invent a conversion rate between them. If an enterprise or regional rate is not published, leave its price unknown and obtain a written quote.

Keep separate estimates for failed attempts, optional tools, storage, exports, human quality review and tax. Use source-verified line items from the companion CSV instead of typing permanent numbers into your forecasting formulas. Where an old plan or model was superseded, preserve its original historical record and create an additive later snapshot. This guide does not claim a product test, performance ranking, independent legal assessment or live procurement approval.

Availability, alternatives and editorial limits

Available plans, supported endpoints, trial eligibility, quota reset and contract rights may vary by region and account even when a public website displays a single global currency. Compare an alternative only with the same accepted task, input/output volume, model version and privacy requirements. Vendor-provided marketing claims are not independent evidence of quality. This historical pricing article may be cross-referenced by the model, media or enterprise procurement team, but its purpose is to explain the billing mechanics without duplicating their evaluations or inventing test scores.