Provider details
Learn more about this provider, then browse the models currently listed under it.
MiniMax (minimax.io)
M3 reaches frontier capability on coding and agentic tasks, introduces the brand-new MSA (MiniMax Sparse Attention) supporting up to 1M context, and is a natively multimodal model. It is the only domestic model combining all three Frontier essentials and will be the only open-source one in this class.
MiniMax (minimax.io)
We look forward to users and developers exploring even more interesting use cases with M2.7. MiniMax Agent: agent.minimax.io. API: platform.
MiniMax (minimax.io)
MiniMax released H3, a new multimodal video-generation model, on 2026-07-31, accepting text, image, video, and audio inputs and capable of producing up to 15-second clips at 2K resolution with native stereo sound. The model also supports editing existing content and transferring motion between videos using mixed-format MiniMax announced plans to publish H3's model weights within days, extending the open-weight trend seen in Chinese language models into video generation, and said the system is designed to run on several Chinese-made chips as part of the sector's effort to reduce dependence on U.S. semiconductors. The company framed H3
MiniMax (minimax.io)
MiniMax is raising approximately $2 billion, with more than half coming from new shares and the rest via convertible bonds, following an earlier ~$619 million Hong Kong listing. Shares fell 9.8% on the news, and the company is separately pursuing a reported $6.5 billion convertible bond offering maturing in 2027 at a c The article embeds several technical claims about M3 that are relevant to developers: 427 billion parameters, 1 million-token prompt length, and vendor-reported 9× prefill and 15× decode speedups over the previous flagship, attributed to MiniMax Sparse Attention — described as a FlashAttention-based technique that redu
MiniMax (minimax.io)
MiniMax M3, released June 1, 2026, is a native multimodal Mixture-of-Experts model with approximately 428 billion total parameters and roughly 23 billion active per token, supporting text, image, and video input with text output. It offers an API context window of up to 1,000,000 tokens (with a stated guaranteed minimu API pricing for M3 is $0.30 per million input tokens and $1.20 per million output tokens for prompts up to 512K tokens, doubling to $0.60/$2.40 per million above that threshold, with MiniMax recommending 131,072 tokens for the M3 reasoning mode. The profile flags that headline benchmark scores are vendor-evaluated and
MiniMax (minimax.io)
Complete list of MiniMax latest updates for June 2026: get every product news, release note, and changelog from MiniMax summarized in one timeline.
MiniMax (minimax.io)
MiniMax has publicly released MiniMax-H3, an open-weights, general-purpose multimodal video model that reads text, images, video, and audio in a unified context and generates coherent audiovisual outputs from any mix of those inputs. Listed capabilities include text-to-video generation, first-and-last-frame workflows, The release is accompanied by Day-0 ecosystem support from fal, Pika, and Civitai, with fal.ai offering a hosted H3 API at pay-per-use pricing (no minimums or subscriptions) and OpenRouter also providing access routes to MiniMax models. Developers can integrate H3 via the fal.ai SDK in Python or JavaScript using an API
What they do
MiniMax develops multimodal AI models across text, speech, video, music, and images and serves them through its API platform on the minimax.io endpoint.
How they were founded
MiniMax says it was founded in early 2022 by Yan Junjie, who previously worked at SenseTime, to build frontier multimodal AI systems.
Models served
These model results are sorted newest first so you can quickly see the latest options from this provider.
