MiniMax M2.7
LLMLLMReasoningReasoningTool-UsingTool-Using
MiniMax M2.7 is a MoE language model with 230B parameters (10B active during inference), released on March 18, 2026.
MiniMax M2.7 is a MoE language model with 230B parameters (10B active during inference), released on March 18, 2026.
Reasoning★
Reasoning
Multi-step reasoning★
Reasoning
Long context★
Reasoning
Coding★
Coding
Function Calling
Planning
Structured output★
Structured gen.
Multilingual★
Language
Planning★
Planning
Streaming output
Reasoning
MiniMax-M2.7: $0.30/M input tokens, $1.20/M output tokens (pay-as-you-go). Prompt caching read: $0.06/M tokens, prompt caching write: $0.375/M tokens. A faster 'highspeed' variant (MiniMax-M2.7-highspeed) is available at $0.60/M input and $2.40/M output.
INPUT
$0.3000 / per 1M tokens
OUTPUT
$1.2000 / per 1M tokens
CACHE
$0.0600 / per 1M tokens
TOTAL
for 10K tokens
MiniMax-M2.7 standard variant. Prompt caching write: $0.375/M tokens.
MiniMax-M2.7-highspeed variant. Same capability, faster inference. Prompt caching write: $0.375/M tokens.