Back to Explore
MiniMax-M3
MiniMax·Released 2026-06-01
1.0M ctxMultimodal
Comparison data ready93% coverage16/16 fields directly observedUpdated Jul 23, 2026, 6:01 AM
About
MiniMax-M3 is a multimodal foundation model from MiniMax. It supports text, image, and video inputs with text output, a 1M-token context window, and is suited for long-horizon agentic work, coding,...
Related Models
Pricing
Input
$0.30
per 1M tokens
Output
$1.20
per 1M tokens
Blended
$0.53
per 1M tokens
Cheaper than 55% of models. Median price is $0.75/1M tokens.
Cost Calculator
Tokens per day1M
100K100M
Daily
$0.53
Monthly
$15.75
vs. Similar Models
DeepSeek V4 Pro (Reasoning, Max Effort)Q:-0.1
$0.54+4%
GPT-5.3 Codex (xhigh)Q:-0.1
$4.81+817%
Kimi K2.6Q:-0.2
$1.71+226%
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Q:-0.7
$10.00+1805%
Performance
100
tokens/sec
Faster than 54% of models
1.32
seconds
Faster than 41% of models
21.25
seconds
Faster than 25% of models
Market Median
95 tok/s
6% faster
Median TTFT
1.10s
20% slower
Throughput/Dollar
191
tok/s per $/1M
Speed Comparison
Qwen3 32B (Non-reasoning)
100 tok/s-1%
Claude 4.5 Haiku (Non-reasoning)
101 tok/s+1%
GPT-5.6 Terra (Non-reasoning)
102 tok/s+1%
Context Window
1.0M
tokens
Larger than 87% of models
Max Output
512K
tokens
49% of context
Benchmarks
MMLU-ProNot evaluated
GPQA Diamond
92.9%
HLE
37.1%
LiveCodeBenchNot evaluated
SciCode
45.4%
TerminalBench Hard
42.4%
MATH-500Not evaluated
AIMENot evaluated
AIME 2025Not evaluated
IFBench
82.9%
Long Context Recall
74.0%
Tau2
88.9%
Market AverageTop Score