Model Detail
MiniMax-M3
—MiniMax-M3 is a code generation model with 213.5B parameters released by MiniMaxAI. The model is registered under the image-text-to-text pipeline tag on Hugging Face, and supports text+image+video->text inputs, distributed under a other license.
MiniMax-M3 reports a Chatbot Arena ELO of 1,448 across 11,264 votes. Other benchmark slots are still empty in our dataset, so this single figure is best read as a partial picture rather than a full evaluation.
MiniMax-M3 is priced at $0.39/M input tokens and $1.56/M output tokens. Operationally the model offers a 524K-token context window, which matters when sizing it for prompt-heavy or latency-sensitive workloads. At this input rate the model sits in the commodity tier and is suitable for high-volume workloads where per-call cost dominates the decision.
MiniMax-M3 ships with 213.5B parameters. Total weight footprint is approximately 427.0 GB, which is the relevant figure when planning local-inference VRAM. Distribution is governed by the other license — review the exact terms before commercial deployment.
MiniMax-M3 is best fit for code completion, repository-scale Q&A, and pair-programming integrations, high-volume batch jobs where per-call cost dominates the budget, and long-context tasks such as full-codebase analysis or book-length summarization (524K tokens). It is a less obvious choice for one-shot generation of security-critical code without review. Treat this as a starting matrix rather than a benchmark verdict — the right deployment usually depends on the specific evaluation suite that mirrors your workload.
Minimax-optimal Scaling Laws in Ridge Regression via Transfer Learning
arXiv:2610.23337v1 Announce Type: new Abstract: Pretrained models are often reused across downstream tasks, motivating the question of whether this transfer learning procedure can provide a substantial performance improvement, namely, in terms of the scaling law. We address this question in the prot
When Is Muon Optimal? Radial Fragility and Minimax Geometry
arXiv:2610.21549v1 Announce Type: cross Abstract: Muon replaces a matrix gradient by an approximate polar factor and has shown strong empirical performance in neural-network training. The Isotropic Curvature Model (ICM) provides a curvature-based explanation for this update by asking when gradient o
Two Sample Sizes in Test-Time LLM Ensembling: Buffered ERM and Minimax Theory
arXiv:2610.21363v1 Announce Type: cross Abstract: Test-time scaling can improve the performance of large language models by generating multiple answers to a question and aggregating them. Weighted ensembles extend this idea by drawing answers from several models, with the weights determining how oft
Discrete Diffusion Models are Minimax Optimal Under Sparsity
arXiv:2610.19020v1 Announce Type: cross Abstract: Discrete diffusion models have emerged as a leading alternative to autoregressive models for generating text and other categorical data, yet their theoretical foundations remain less developed than those of their continuous counterparts. For continuo
Minimax estimation of the pseudo-spectral gap of continuous Markov chains, with applications to empirical concentration bounds
arXiv:2610.20007v1 Announce Type: cross Abstract: We study the minimax estimation of the pseudo-spectral gap $\gamma_{ps}$ of a Markov chain from the observation of a single trajectory. Extending earlier works on finite-state space chains by Wolfer and Kontorovich (2024), we provide an estimator in
Variance-Reduced Policy Gradient is Nearly Minimax-Optimal
arXiv:2610.13692v1 Announce Type: cross Abstract: Standard Policy Gradient (PG) algorithms suffer from high estimation variance because they rely on Monte Carlo updates. While off-policy variance reduction mitigates this problem, establishing optimal sample complexity currently requires either restr
