MiniMax-M3
AvailableA MiniMax model with a 1M-token context window, image input and tool calling. On the MiniMax Open Platform API since June 1, 2026.
Access details: MiniMax Open Platform API
An alias. The provider can point it at a newer version.
MiniMax-M3Open weights on Hugging Face: MiniMaxAI/MiniMax-M3
Updated Oct 1, 2026
Capabilities & limits
- Context window
- 1M
- Maximum output
- 512K
- Image input
- Supported
- Tool calling
- Supported
- Structured output
- Unsupported
Pricing
USD / 1M tokens- Input
- $0.30
- Output
- $1.20
Standard rates per million tokens. Discounts and long-context rates, where offered, are on MiniMax's pricing page.
Full rate card 12 rates
All prices in USD.
Standard input, up to 512k input tokens
$0.30 per 1M tokens- standard tier
- up to 512k input tokens
- current price after permanent 50% off (list $0.60)
Standard output, up to 512k input tokens
$1.20 per 1M tokens- standard tier
- up to 512k input tokens
- current price after permanent 50% off (list $2.40)
Standard cache read, up to 512k input tokens
$0.06 per 1M tokens- standard tier
- prompt caching
- up to 512k input tokens
- current price after permanent 50% off (list $0.12)
Standard input, over 512k input tokens
$0.60 per 1M tokens- standard tier
- over 512k input tokens
- current price after permanent 50% off (list $1.20)
Standard output, over 512k input tokens
$2.40 per 1M tokens- standard tier
- over 512k input tokens
- current price after permanent 50% off (list $4.80)
Standard cache read, over 512k input tokens
$0.12 per 1M tokens- standard tier
- prompt caching
- over 512k input tokens
- current price after permanent 50% off (list $0.24)
Priority input, up to 512k input tokens
$0.45 per 1M tokens- priority tier (service_tier priority)
- up to 512k input tokens
- current price after permanent 50% off (list $0.90)
Priority output, up to 512k input tokens
$1.80 per 1M tokens- priority tier (service_tier priority)
- up to 512k input tokens
- current price after permanent 50% off (list $3.60)
Priority cache read, up to 512k input tokens
$0.09 per 1M tokens- priority tier (service_tier priority)
- prompt caching
- up to 512k input tokens
- current price after permanent 50% off (list $0.18)
Priority input, over 512k input tokens
$0.90 per 1M tokens- priority tier (service_tier priority)
- over 512k input tokens
- current price after permanent 50% off (list $1.80)
Priority output, over 512k input tokens
$3.60 per 1M tokens- priority tier (service_tier priority)
- over 512k input tokens
- current price after permanent 50% off (list $7.20)
Priority cache read, over 512k input tokens
$0.18 per 1M tokens- priority tier (service_tier priority)
- prompt caching
- over 512k input tokens
- current price after permanent 50% off (list $0.36)
Pay-as-you-go USD. The price list shows a struck-through list price with a permanent 50% discount; the discounted figure is recorded as the current rate. Priority is 1.5x standard per the page. No cache-write rate is listed for M3.
Price history
Oct 1, 2026
Input $0.30 per million tokens
Oct 1, 2026
Output $1.20 per million tokens
Technical details
- Parameters
- ~428B
- Active parameters
- ~23B
- Architecture
- Mixture-of-experts with MiniMax Sparse Attention (MSA)
- Training / knowledge cutoff
- Not published
- Input modalities
- Text, image, video
- Output modalities
- Text
From MiniMax
Official announcement ↗How they describe it
Coding & Agentic Frontier. 1M-context MSA. Native Multimodality
What they say is new
The first open-weight model with three frontier capabilities.
Release & lifecycle
- First available
- Jun 1, 2026
- available
- As of Oct 1, 2026