minimax

The launch of MINIMAX contract trading on the Hyperliquid HIP-3 ecosystem has garnered significant attention, with approximately $480,000 in long positions and a current price of $60.7 for $MINIMAX .

Entry: 60.7 🔥
Target: 70 🚀
Stop Loss: 55 ⚠️

This development is notable, especially considering the recent market speculation and pressure on MiniMax's stock price. The price has dropped by around 53% in the first two weeks of June due to concerns about a potential increase in the floating supply and a rating downgrade by Morgan Stanley.

Not financial advice. Manage your risk.

#MINIMAX #LongSetup #ContractTrading
❌

On the first day of MINIMAX's on-chain contract launch, a trader went short with 10x leverage.

Today, Hyperliquid HIP-3 ecosystem has launched MINIMAX (MiniMax) contract trading. As of this writing, approximately $480,000 in positions have been opened for MINIMAX. A trader shorted MINIMAX using 10x leverage as the stock price faces pressure from the unlocking expectations. The on-chain contract market's reaction to AI concept stocks is accelerating.

Why it matters: On-chain contracts for AI concept stocks provide Web3 users with a new trading dimension, but they also bring higher volatility risk.

#Hyperliquid #MINIMAX #AI #contract trading

MiniMax was 'forced' by devs to drop prices by 50% permanently, AI model price wars heating up

Just two weeks post-launch, the flagship model MiniMax-M3 on the MiniMax open platform has undergone a dramatic price adjustment. After a collective backlash from developers against a stealth price hike, MiniMax was compelled to offer a 'permanent 50% discount': input prices dropped from $0.60 to $0.30 per million tokens, and output prices fell from $2.40 to $1.20. The pricing power in the AI model market is shifting from vendors to the developer community.

Why it matters: The trend towards transparent AI model pricing is unstoppable, and the collective bargaining power of the developer community is reshaping the profit distribution framework in the AI supply chain.

#MiniMax #AI #大模型 #pricewar

MiniMax M3 Officially Open-Sourced: 428 Billion Parameters Native Multimodal Model, Supports Million Contexts

The domestic large model provider MiniMax has officially open-sourced their native multimodal mixture of experts (MoE) model, MiniMax M3, on Hugging Face. It boasts a total of 428 billion parameters, with 23 billion parameters activated per token, natively supporting 1 million ultra-long contexts. The team also released an MXFP8 quantized version, compatible with mainstream frameworks like SGLang and vLLM. The model supports both Thinking/Non-thinking dual inference modes and achieves joint training of text, images, and video during the pre-training phase.

Why it Matters: This marks the first time a domestic large model has simultaneously reached SOTA levels in native multimodal capabilities, million ultra-long contexts, and open-source weights across three dimensions, significantly lowering the barrier for developers to integrate complex multimodal capabilities into AI applications.

#MiniMax #AI #开源 #大模型 #multimodal

MiniMax open-source math proof framework MaxProof: Using evolutionary search to push M3 to the IMO gold standard

MiniMax founder Yan Junjie announced the open-source math proof framework MaxProof. This framework automatically selects the right difficulty training data through evolutionary search, eliminating the need for manual labeling, ultimately pushing MiniMax M3 to the international math olympiad gold benchmark. This is a significant open-source contribution from the Chinese AI team in the field of mathematical reasoning, following DeepSeek.

Why it matters: MaxProof addresses the pain points of high training data labeling costs and difficulty matching in AI mathematical reasoning, and it has the potential to accelerate breakthroughs in AI applications in scientific research and complex reasoning.

#AI #MiniMax #大模型 #artificialintelligence

MiniMax officially launches the M3 model: coding and agent tasks reach cutting-edge levels

MiniMax has rolled out the MiniMax M3 model, featuring the brand-new MSA (MiniMax Sparse Attention) architecture, supporting up to 1M ultra-long context. This multimodal model can handle image and video inputs as well as desktop operations. It achieves cutting-edge performance in specialized tasks like coding and agents.

Why it matters: The 1M context and multimodal capabilities of MiniMax M3 signify a continued breakthrough by Chinese AI teams in the general AI space, with the MSA sparse attention architecture providing a new technical pathway for long-context reasoning.

#MiniMax #AI #人工智能 #大模型 #Web3