Skip to content

InferenceMiniMax

MiniMax

Multimodal foundation models and developer API for text, code, video, speech, and music.

Categories
InferenceVideo
Pricing
FREEMIUM
Source
Proprietary
Hosting
Cloud
Platforms
WebAPI
Models
Self-contained (on-device)
Verified
Jun 14, 2026

MiniMax is a Shanghai foundation-model lab whose platform serves its own model family through a developer API and agent app: the M-series LLMs (M2/M3) built for coding and agentic workflows with up to a 1M-token context, the Hailuo video models, and MiniMax Speech and Music. Developers get chat completions, text-to-speech, and text-to-video on token-based pricing, with a free agent tier for getting started.

Pros & cons

  • Frontier coding/agent M-series models
  • Up to 1M-token context
  • Open-weight models on Hugging Face
  • Multimodal: text, video, speech, music
  • Aggressive token pricing
  • China-based; data-residency considerations
  • Free agent tier is credit-limited
  • Docs less mature than US incumbents

Tags

Further reading

Alternatives to MiniMax (4)

View all Inference →
  • View DeepSeek details
    AssistantFREEMIUM

    DeepSeek

    DeepSeek

    Open, low-cost chat with strong reasoning. Free to use.

    DeepSeek's assistant — chat with a reasoning mode and web search, backed by the open-weight DeepSeek models that reset the cost curve for frontier-grade quality.

    Worth knowing

    Spun out of Chinese quant hedge fund High-Flyer in 2023, funded by it rather than by venture capital.

    • chat
    • assistant
    • reasoning
    • open-weights
    Open
  • View Z.ai details
    AssistantFREEMIUM

    Z.ai

    Z.ai (Zhipu AI)

    Zhipu's GLM assistant — frontier open-weight chat, free to use.

    Z.ai is the international assistant of Zhipu AI, built on the GLM model family. The free chat runs the flagship GLM-5 and GLM-5.1 models with reasoning and agentic modes, and the same models are served to developers over a low-cost, OpenAI-compatible API. GLM-5's open weights ship under the MIT license.

    Worth knowing

    Maker Zhipu's January 2026 Hong Kong IPO ($558M at a $7.1B valuation) made it the world's first publicly traded foundation-model company.

    • chat
    • assistant
    • open-weights
    • reasoning
    • +1
    Open
  • View Together AI details
    InferenceFREEMIUM

    Together AI

    Together

    Fine-tuning + inference for open-weights models. Broad coverage.

    Hosted inference and fine-tuning across hundreds of open-weights models (Llama, Mistral, DeepSeek, Qwen, etc.). Strong pricing for inference-at-scale; LoRA + full fine-tuning supported.

    Worth knowing

    Co-founded by Stanford's Percy Liang and FlashAttention author Tri Dao; raised $305M at a $3.3B valuation.

    • inference
    • fine-tuning
    • open-weights
    • lora
    Open
  • View Hailuo AI details
    VideoFREEMIUM

    Hailuo AI

    MiniMax

    Text- and image-to-video generation from MiniMax.

    MiniMax's consumer video generator, turning text prompts and reference images into short cinematic clips with subject-reference for consistent characters. Available on the web and as iOS and Android apps. A free tier offers limited credits; subscriptions add HD output, faster generation, and commercial use.

    Worth knowing

    Maker MiniMax (Alibaba- and Tencent-backed) IPO'd in Hong Kong in Jan 2026, its stock roughly doubling on debut to a ~$13.7B valuation.

    • video-gen
    • text-to-video
    • image-to-video
    • minimax
    Open
Open MiniMax