MiniMax launched the Token Plan on March 23, 2026, a flat-rate API subscription that gives developers access to video, music, speech, image, and text generation models under a single key. It is the first all-modality API subscription from a major AI provider, replacing MiniMax's original Coding Plan with a unified offering that covers every creative modality the company ships.

For the broader landscape, see our complete guide to AI video generation in 2026.

What Happened

MiniMax upgraded its existing Coding Plan into the Token Plan, bundling the company's full model suite into one subscription. The plan includes access to MiniMax's M2.7 text model, Hailuo video generation, Speech voice synthesis, a music generation model, and image generation tools. Developers get one API key and one predictable bill instead of juggling separate accounts and pay-per-token charges across different modalities.

The plan uses 5-hour measurement cycles for programming model usage and adds independent multi-modal quotas for Plus tier and above. MiniMax says the Token Plan saves approximately 20% compared to pay-as-you-go pricing. Dedicated resource packages are also available for the Speech 2.8 and Hailuo 2.3 models.

Why It Matters

Creators building multi-modal AI workflows currently sign up for separate services to handle video, audio, image, and text generation. Each service has its own pricing model, API key, and billing cycle. The Token Plan collapses that into a single subscription, which matters most for developers shipping products that combine multiple modalities, like AI video editors that need voice synthesis, background music, and thumbnail generation in the same pipeline.

This also signals a shift in how AI companies package their products. Instead of competing on individual model benchmarks, MiniMax is competing on breadth of access. The M2.7 model already ranks as one of the strongest 10B-parameter agent models, and Hailuo remains a popular choice for AI video generation. Bundling them under one plan could pull developers away from mixing providers.

Key Details

  • Models included: M2.7 (text/code), Hailuo (video), Speech (voice), Music (audio), Image (generation)
  • Pricing: Flat-rate subscription tiers, approximately 20% savings vs. pay-as-you-go
  • Usage model: 5-hour cycles for text/code, independent quotas for multi-modal features on Plus plans
  • Traffic note: Dynamic traffic controls apply during peak weekday hours; high-concurrency tasks may need pay-as-you-go mode
  • Availability: Live now at platform.minimax.io

What to Do Next

If you are already using Hailuo for video or MiniMax for text, check whether the Token Plan covers your usage at a lower cost than your current pay-as-you-go spending. The bundled pricing makes the most sense for workflows that touch at least two modalities. Developers running single-modality projects at high volume should compare the flat-rate quotas against their actual token consumption before switching.