MiniMax

MiniMax M3

MiniMax

MiniMax M3 is a native multimodal MoE model trained jointly on text, images, and video. It adds MiniMax Sparse Attention for efficient million-token context and targets long-horizon coding and knowledge-work agents. The model supports enabled, adaptive, and disabled reasoning modes. MiniMax launched hosted API and Agent access on June 1 and later released downloadable weights. The full checkpoint has about 428B parameters with approximately 23B activated. Its MiniMax Community License permits non-commercial use and imposes additional conditions for commercial deployment.

MiniMax M2.7

MiniMax

MiniMax M2.7 improves professional software engineering, office work, and character consistency, and explores model participation in its own training and harness improvement. It is a text MoE model in the M2 family, exposed through MiniMax API and Agent products. Downloadable weights were released after the initial hosted launch. The current weight license permits non-commercial use; commercial weight deployment requires prior written authorization. Agent Team workflows are orchestration around the model, not additional model modalities.

MiniMax M2.5

MiniMax

MiniMax M2.5 extends the M2 family through reinforcement learning in complex real-world environments. It targets software engineering, search and tool use, and office work, with improved planning before code generation and stronger professional web research. The release provides public modified-MIT weights and official API/Agent access. A high-speed hosted service variant shares the model identity and is grouped here. Reported research results using context management depend on that agent implementation.

MiniMax-M2-her

MiniMax

MiniMax-M2-her is MiniMax’s language-model specialization for immersive role-play and sustained multi-turn conversation. Training focuses on maintaining character and world consistency, advancing narrative structure, and adapting to users’ implicit preferences. A dedicated Text Chat API defines role-oriented inputs, text-only interaction, a 64K context, and a 2,048-token completion limit. The named model was available during January 2026 before the January 27 technical article; the precise first public day is not established here. Downloadable weights and a checkpoint-level architecture are not verified.

MiniMax M2.1

MiniMax

Enhanced version of MiniMax-M2 with significantly improved multi-language programming capabilities, built for real-world complex coding tasks. Features exceptional performance in WebDev, AppDev (Android/iOS native development), and office automation scenarios. Delivers more concise and efficient responses compared to M2 while maintaining powerful general intelligence. Demonstrates excellent generalization across various programming tools and Agent frameworks.

MiniMax M2

MiniMax

MiniMax M2 is an efficiency-focused MoE language model for coding and agent workflows. It activates about 10B parameters from roughly 230B total, targeting multi-file editing, iterative run-and-fix loops, and test-validated repairs. Open weights complement MiniMax API and Agent access. Interleaved reasoning and tool invocation support end-to-end developer tasks rather than isolated code completion. Weight access uses modified MIT terms, which should not be confused with an unmodified MIT license.

MiniMax M1

MiniMax

MiniMax M1 is a hybrid-attention MoE reasoning model derived from MiniMax-Text-01. It combines lightning attention with softmax attention and uses CISPO reinforcement learning to scale long reasoning. A million-token input context supports large documents and repository-level agent tasks. The 40K and 80K thinking-budget checkpoints are represented together as variants, with Apache-2.0 weights available for self-hosting.

MiniMax-Text-01

MiniMax

MiniMax-Text-01 is a long-context language foundation model combining lightning attention, softmax attention, and a sparse MoE backbone. It has 456B total parameters and 45.9B active per token; the research demonstrates context extension up to four million tokens. Public weights and reference inference code support long-document understanding and general language tasks. Model weights use the MiniMax Model License, separate from the MIT license on code. Historical launch/API claims should not be interpreted as guarantees that the original endpoint still serves this checkpoint.

MiniMax-VL-01

MiniMax

MiniMax-VL-01 adds a 303M-parameter vision encoder and MLP projector to MiniMax-Text-01. The model supports image-grounded conversation, document understanding, and visual reasoning while inheriting the hybrid-attention language backbone. It was trained in stages on image-caption and instruction data, with public weights and reference inference code available. Model weights use the MiniMax Model License, separate from the MIT license on code. Historical launch/API claims should not be interpreted as guarantees that the original endpoint still serves this checkpoint.