AIHubMix Blog

Announcements, tutorials, model analysis, and product updates from AIHubMix.

• 

  Migrating to Kimi K3.1 Pro: Fix the 400s, Mind the Max-Effort Default, and Don't Mistype Highspeed

  Kimi K3.1 Pro is live in Kimi Code, on Moonshot's API as kimi-k3.1-pro, and on AIHubMix under the same ID. Moonshot gives it the same parameter constraints as Kimi K3, so code that already runs on K3 should move with a model-ID change and a fresh session. Code written for K2.x now has a deadline on Moonshot's platform: Moonshot closes the K2 series to newly registered users there on 31 October 2026 and plans to retire it by the end of December. That code is also the most likely to break, because

  October 23, 202618 min readTutorial
• 

  Kimi K3.1 Pro vs K3: What Changed, What It Costs, and Who Should Switch

  Kimi K3.1 Pro is the first update to Kimi K3, the 2.8-trillion-parameter open-weight model Moonshot AI shipped on 16 July 2026. It launched on 23–24 October 2026 in the Kimi app, Kimi Code and the Kimi API. Moonshot's launch post calls it "the first model in our new K3.1 series and our most capable 3T-class open-weight model to date": a post-trained version of K3, trained with much more post-training compute across "hundreds of downstream domains". Moonshot claims gains in coding, 3D creation, a

  October 23, 202614 min readNews
• 

  GLM-5.3 Prime vs GLM-5.3: Is 2× the Price Worth the Speed?

  GLM-5.3 Prime is not a new model. It runs the same GLM-5.3 weights in a faster serving mode: Alibaba Cloud Model Studio's Prime mode, which promises 1.5 to 2 times the output speed of the standard API. The answers should be the same quality. What you pay extra for is tokens per second. On AIHubMix, glm-5.3-prime costs $2.35 per million input tokens and $8.23 per million output tokens, about 2.1× what GLM-5.3 costs today. The short answer: * Worth it when a person or an agent loop sits waiti

  October 21, 202610 min readOpinion
• 

  AI Halloween Photo Templates: 12 Looks for Costumes, Pets, Shops and Scenes

  12 ready-made AI Halloween photo looks on AIHubMix for costumes, pets, shops and no-photo scenes. Pick one, open it in the Playground with the prompt filled in, and add your photo.

  October 20, 20265 min readAnnouncement
• 

  Claude Code settings.json and Environment Variables: Base URL, Auth Token and Model

  Claude Code takes its configuration from JSON settings files and from environment variables. To run it through a gateway instead of Anthropic's API, you need two variables, ANTHROPIC_BASE_URL and a token, and usually a model. The variables are simple. What confuses people is where to put them, because the same variable can sit in your shell profile, in several settings.json files and in VS Code's settings. This guide starts with a setup that works, then explains where the files live, which val

  October 20, 20267 min readTutorial
• 

  MiniMax M3.1 Flash vs DeepSeek V4.1 Flash vs Claude Haiku 5.5: One Bug Fix, Three Bills

  Three cheap coding models shipped within four weeks of each other: DeepSeek V4.1 Flash (with its 1017 update), Claude Haiku 5.5, and MiniMax M3.1 Flash, which spent its first week as the anonymous Space Bunny. All three have a 1M-token context window and cost between $0.10 and $0.17 per million input tokens. So which one should you use for routine coding? AIHubMix gave all three the same bug fix, three times each, through one API key. Each model fixed the bug correctly every time, but the bill

  October 19, 20267 min readOpinion
• 

  MiniMax M3.1 Flash Effort Levels: Why the Default Cost 7× Low on the Same Bug Fix

  MiniMax M3.1 Flash has five thinking levels and no off switch. If you don't pick a level, the model picks the deepest one for you. MiniMax's text generation guide says effort "defaults to max when omitted." That default is expensive in a way that's easy to miss. AIHubMix ran the same two-bug fix at every level. The default used about 10× the thinking tokens of low and took about 6× as long, and every run at every level passed all 215 tests. The model's thinking cost rose with the effort settin

  October 19, 20267 min readTutorial
• 

  Moving Off MiniMax M2.5/M2.7 Before Nov 18: Fixing M3.1 Flash's "requires adaptive thinking" 400

  MiniMax is retiring M2.5 and M2.7 on 18 November 2026, and the switch has already started. Since 19 October, the two older families have been closed to new users, and plan requests that name them are served by M3.1 Flash. If you're on pay-as-you-go, you have four weeks to change the model ID yourself. Most M2.7 code runs on M3.1 Flash unchanged. The trouble comes from one design choice: M3.1 Flash always thinks, and it thinks at the deepest level unless you say otherwise. That produces one lou

  October 19, 20269 min readTutorial
• 

  MiniMax M3.1 Flash Is Space Bunny: What Changed from M2.7 and M3, and What It Costs

  In late September, a model with no maker's name on it topped OpenRouter's daily chart, ahead of DeepSeek V4.1 Flash, and stayed OpenCode's most-used model for five days straight. It was listed as Space Bunny Alpha, and it was free. That model was MiniMax M3.1 Flash. As of 19 October 2026 you can call it per token on AIHubMix as minimax-m3.1-flash, at $0.165 per million input tokens and $0.66 per million output tokens. The short version: * It replaces M2.7. Starting 19 October, MiniMax sends

  October 19, 20268 min readNews
• 

  How to Set Up Codex config.toml: Custom API, Proxy, and Common Errors

  Codex keeps its settings in one file, config.toml: which model to use, which API to call, and how much the agent may do without asking. The format changed several times in 2026, and many guides from earlier in the year show settings that now stop Codex from starting. This guide gives you a working config first, then the parts most people change: profiles, project settings, the sandbox, and a proxy. At the end is a list of the errors you're likely to see, with what each one means. Checked again

  October 19, 20267 min readTutorial