MiniMax M3.1 Flash Launched: 428B MoE Model at $0.15/M Input
MiniMax released M3.1 Flash, a 428B MoE model with 23B active parameters, a 1M-token context window, and API pricing at $0.15 per million input tokens.
Stock photo for illustration only, not from the actual event
- MiniMax has officially released its new M3.1 Flash AI model featuring 428 billion total parameters.
- Utilizes a Mixture of Experts architecture with roughly 23 billion active parameters and a 1M-token context.
- Offers competitive API pricing of $0.15 for input and $0.60 for output per one million tokens.
- Model weights are scheduled to be released to the public on October 23.
The artificial intelligence landscape welcomes a new contender as MiniMax rolls out its M3.1 Flash model, engineered to deliver robust agentic capabilities combined with high-throughput processing efficiency.
This latest release boasts an expansive scale of 428 billion total parameters. By leveraging a Mixture of Experts (MoE) architecture, the model activates approximately 23 billion parameters per run, striking a balance between high intelligence and operational speed.
The introduction of high-parameter MoE models at heavily discounted API rates highlights the ongoing market trend toward cost-efficiency in enterprise AI deployment. A 1-million-token context window is particularly crucial for agentic workflows, enabling autonomous systems to parse entire codebases or comprehensive documentation archives in a single session.
Stock photo for illustration only, not from the actual event
Furthermore, M3.1 Flash supports an impressive 1-million-token context window, giving developers the capacity to handle extensive prompt inputs and complex multi-step reasoning tasks without truncation issues.
API access is priced economically at $0.15 per million input tokens and $0.60 per million output tokens. Developers looking to host or inspect the architecture directly will also note that the model weights are slated for release on October 23.
Source: MarkTechPost
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment