About
MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digital working environments, M2.5 builds upon the coding expertise of M2.1...
Related Models
Pricing
Input
$0.12
per 1M tokens
Output
$0.48
per 1M tokens
Blended
$0.21
per 1M tokens
Cheaper than 68% of models. Median price is $0.54/1M tokens.
Cost Calculator
Tokens per day1M
100K100M
Daily
$0.21
Monthly
$6.30
vs. Similar Models
Qwen: Qwen3.5 397B A17BQ:0.0
$0.90+329%
Claude 4.1 Opus (Reasoning)Q:0.0
$30.00+14186%
GPT-5 (medium)Q:0.0
$3.44+1537%
Qwen: Qwen3.5-27BQ:+0.1
$0.54+155%
Performance
175
tokens/sec
Faster than 81% of models
6.57
seconds
Faster than 14% of models
17.99
seconds
Faster than 28% of models
Market Median
94 tok/s
86% faster
Median TTFT
1.11s
490% slower
Throughput/Dollar
834
tok/s per $/1M
Speed Comparison
Qwen3.5 122B A10B (Non-reasoning)
174 tok/s-0%
Nova Lite
176 tok/s+1%
Command A+
174 tok/s-1%
Context Window
205K
tokens
Larger than 57% of models
Max Output
197K
tokens
96% of context
Benchmarks
MMLU-ProNot evaluated
GPQA Diamond
84.8%
HLE
19.1%
LiveCodeBenchNot evaluated
SciCode
42.6%
TerminalBench Hard
34.8%
MATH-500Not evaluated
AIMENot evaluated
AIME 2025Not evaluated
IFBench
71.6%
Long Context Recall
66.0%
Tau2
95.3%
Market AverageTop Score