Skip to content
DigIntoAI/Model intelligence
Model intelligence feed

Model intelligence.

Follow the changes. Understand what comes next.

Editor’s pickMiniMax
MiniMax/MiniMax-H3-Image

MiniMax Releases MiniMax-H3-Image, an Open-Weight Image Model Available via API, MiniMax Code, and Hailuo AI

MiniMax released MiniMax-H3-Image on November 10, 2026 (Beijing time), with open weights and access through its API, MiniMax Code and Hailuo AI. MiniMax claims 4096×4096 generation in 10 seconds and support for up to 16 reference images. Weights use a community license with geographic and commercial restrictions. Pricing is token-based; requests delivering at least 16,600,000 pixels receive 50% off the total cost until November 18, 2026, 00:00 Beijing time, exclusive. Official latency and cost charts are included.

Latest stories10

Newest published
01

MiniMax Releases MiniMax-H3-Image, an Open-Weight Image Model Available via API, MiniMax Code, and Hailuo AI

MiniMax released MiniMax-H3-Image on November 10, 2026 (Beijing time), with open weights and access through its API, MiniMax Code and Hailuo AI. MiniMax claims 4096×4096 generation in 10 seconds and support for up to 16 reference images. Weights use a community license with geographic and commercial restrictions. Pricing is token-based; requests delivering at least 16,600,000 pixels receive 50% off the total cost until November 18, 2026, 00:00 Beijing time, exclusive. Official latency and cost charts are included.

MiniMax-H3-ImageRead story
02

Pika H3 Flex Is Open-Weight—What Actually Runs Locally and What Still Needs the Cloud

Pika released H3 Flex's adapter weights and inference code, post-trained on MiniMax H3. Local generation tops out at 480p/768p; 4K, audio stems, and prompt enhancement stay on the paid API.

Pika H3 Flex Read story
03

AI Says the Code Is Done. Who Checks It? MiniMax Just Published Its Team's Review Playbook

MiniMax open-sourced the workflows its Code team actually uses, including an independent-review skill that has a second AI check the diff with zero access to the author's reasoning. Here's what that fixes.

MiniMaxRead story
04

Same Open Model, Different API: Why Speed and Cost Both Change

Vals AI’s Inference Provider Index (page Updated 10/20/2026; Updates dated 10/22/2026) runs the same open-weight models—DeepSeek V4.1 Flash, Kimi K3, and MiniMax M3—on first-party APIs and hosts such as Together AI, Fireworks, and Baseten. Across 165 agent tasks, overall accuracy stayed within error of the first-party API. Cost per task rose by as much as about 2.3× by host, and for all three models the creator’s API was neither the cheapest nor the fastest. Figures below are from Vals AI, for reference only.

Inference Provider IndexRead story
05

FastH3 Omni on a Local GPU — Thresholds, Speed Caveats, License Limits

FastVideo and Nuva Lab open-sourced FastH3 Omni for reference-guided video with synced audio. This guide maps official local GPU recipes (DGX Spark, RTX 5090, FP8/4090/ComfyUI), keeps the 5.6–11.9× claim tied to eight GB200 matched tests, and spells out the MiniMax-H3 Community License exclusions. We have not measured quality or speed.

FastH3 OmniRead story
06

MiniMax M3.1 Flash Open Weights Are on Hugging Face

As of 08:59 Beijing time on October 23, 2026, open weights for MiniMax-M3.1-Flash are on Hugging Face. The model card describes about 428B total MoE parameters with 23B active per token, multimodal inputs, and up to 1M context under the MiniMax Community License. Pricing and API context remain in the main article.

MiniMax M3.1 FlashRead story
07

What MiniMax Code Mini App Is and Where It Won't Run

MiniMax Code Mini App is a long-running plugin with a frontend, a Node backend that can keep data, and two-way calls with the Agent. It runs only in local macOS and Windows desktop sessions—not in cloud, SSH, or Remote Control. We have not tested it, and Mini App plan fees are not verified from the public pages.

MiniMax Code Mini AppRead story
08

MiniMax Releases M3.1 Flash at $0.15 per Million Input Tokens

MiniMax released M3.1 Flash at $0.15 input and $0.60 output per 1M tokens and claims 73.7% on DeepSWE v1.1. It said on X that weights arrive October 23.

MiniMaxRead story
09

How to Use GLM, Qwen, MiniMax and Other Chinese Coding Plans Inside Claude Code

Keep Claude Code, swap the model: step-by-step setup for Z.ai GLM, Alibaba Cloud Qwen, MiniMax, Kimi, BytePlus and StepFun coding plans, taken from each vendor's official docs. Plus how to switch back to Claude, the errors people hit, and the plan rules worth knowing.

Claude Code / GLM / Qwen / MiniMax M Plan / Kimi Code / BytePlus ModelArk / Step PlanRead story
10

AI coding plans compared: prices, usage limits and buying in Japan

Compare coding subscriptions by monthly price, included usage and extra charges. The November 20 review covers Qwen, Copilot, Cursor, Codex and Claude Code, with Japanese billing notes and the October edition.

Codex / Claude Code / Antigravity / Cursor / GitHub Copilot / GLM / Kimi / Qwen / MiniMaxRead story