MiniMax introduces Token Plan for flat-rate text, speech, music, video, and image APIs
MiniMax introduced a flat-rate Token Plan that covers text, speech, music, video, and image APIs under one subscription. It gives teams one predictable bill across modalities and can be used in third-party harnesses, not just MiniMax apps.
TL;DR
- MiniMax launched a Token Plan that bundles its text, speech, music, video, and image APIs into one flat-rate subscription, which the company calls the "first all-modality API subscription" launch thread.
- MiniMax says the new plan keeps "pricing and M2.7 five-hour usage" the same as its existing Coding Plan while extending access across more modalities plan details.
- The company is pitching the plan around billing predictability: "one key" and "one predictable bill" instead of separate modality-specific charges launch thread.
- MiniMax also says Token Plan usage is "not restricted to specific websites and applications" and can be used in third-party harnesses, which matters for teams wiring it into their own tooling stacks harness support.
What actually shipped?
MiniMax's launch post says the Token Plan is a single subscription for "text, speech, music, video, and image" APIs rather than separate usage-based billing across modalities launch thread. That makes this more of a packaging and procurement change than a new model release, but it is still an implementation-level update for teams that have been mixing generation modes inside one product.
The attached [img:0|pricing table] shows the plan structure spans M2.7 text requests plus daily allowances for Image 01, Speech 2.8, Music 2.5, and Hailuo 2.3 video generation at multiple tiers. MiniMax's plan details also points developers to a subscription page via the Token Plan page and to an open multimodal toolkit via the ClawHub toolkit.
Where can engineers use it, and what are the limits?
MiniMax says the Token Plan is "not restricted to specific websites and applications" and works in "favorite third-party harnesses" harness support. For engineering teams, that is the key operational detail: the subscription is being positioned as API capacity that can travel into external agents and dev tools, not just MiniMax-owned surfaces.
MiniMax also says "pricing and M2.7 five-hour usage remain the same as the Coding Plan" plan details. According to the pricing table post, the entry M2.7 tier starts at 1,500 requests per five hours and scales to 30,000 requests per five hours on the highest tier, while non-text modalities are capped by per-day quotas such as images, speech characters, songs, and short video generations.