MiniMax: MiniMax-M2.1-lightning API
MiniMax-M2.1-lightning is the ultra-accelerated version of the M2.1 series. Its defining innovation is "Lightning Attention" (a linear-complexity modeling approach), which bypasses the "quadratic wall" of traditional Transformers. It delivers approximately 60% faster throughput (reaching ~100 TPS) while maintaining parity in reasoning quality. Notably, it supports an unprecedented 4-million-token native context window, enabling developers to process entire code repositories or massive document sets without lossy RAG pipelines. It is the premier choice for high-speed agentic reasoning and long-context analysis.
- Context window: 204,800 tokens
- Max output: 131,072 tokens
- Input: text
- Output: text
- Reasoning: Supported
- Tool calling: Supported
- Released: 2025-12-23
- Knowledge cutoff: 2025-05-31
Frequently Asked Questions
What is the context window of MiniMax-M2.1-lightning?
MiniMax-M2.1-lightning supports a context window of up to 204,800 tokens.
Does MiniMax-M2.1-lightning support function calling?
Yes. MiniMax-M2.1-lightning supports tool / function calling.
Does MiniMax-M2.1-lightning support reasoning?
Yes. MiniMax-M2.1-lightning is a reasoning-capable model.
What is the knowledge cutoff of MiniMax-M2.1-lightning?
The knowledge cutoff of MiniMax-M2.1-lightning is 2025-05-31.