feat: add MiniMax as a new LLM provider

[octo-patch]

Summary

• Add MiniMax as a new LLM provider with OpenAI-compatible API support
• Register MiniMax-M2.5 (204K context) and MiniMax-M2.5-highspeed models
• Include GUI model selection, provider configuration, and documentation

Changes

New Files

• core/llm/llms/MiniMax.ts — Provider class extending OpenAI with:
  • Temperature clamping to (0, 1] range (MiniMax rejects 0)
  • response_format removal (unsupported)
  • Default API base: https://api.minimax.io/v1/
• packages/llm-info/src/providers/minimax.ts — Model metadata
• docs/customize/model-providers/more/minimax.mdx — Provider documentation

Modified Files

• core/llm/llms/index.ts — Register in LLMClasses
• packages/openai-adapters/src/index.ts — Register OpenAI-compatible adapter
• packages/config-types/src/index.ts — Add "minimax" to provider enum
• packages/llm-info/src/index.ts — Add to allModelProviders
• gui/src/pages/AddNewModel/configs/models.ts — Model entries
• gui/src/pages/AddNewModel/configs/providers.ts — Provider config with API key setup
• docs/customize/model-providers/overview.mdx — Add to hosted services table

Models

Model	Context	Max Output	Description
MiniMax-M2.5	204,800	192,000	Peak performance, complex reasoning
MiniMax-M2.5-highspeed	204,800	192,000	Same performance, lower latency

Test Plan

• Verified MiniMax API connectivity and response format
• Verified temperature constraint handling
• Code follows existing provider patterns (Groq, DeepSeek, Mistral)

---

Summary by cubic

Add MiniMax as an OpenAI-compatible LLM provider with GUI support and docs. Includes MiniMax-M2.5 and MiniMax-M2.5-highspeed models (204K context) and normalizes unsupported params.

• New Features

  • New minimax provider using https://api.minimax.io/v1/ via the OpenAI-compatible adapter.
  • Registers provider and models in LLMClasses, config types, and llm-info.
  • GUI: provider setup with API key and model entries for MiniMax-M2.5 and MiniMax-M2.5-highspeed.
  • Request handling: clamps temperature to (0, 1] and removes unsupported response_format.
  • Docs: provider guide and overview entry.

• Migration

  • Configure provider: minimax with your API key, or set MINIMAX_API_KEY.
  • For China, set apiBase to https://api.minimaxi.com/v1/.

^{Written for commit a15305e. Summary will update on new commits.}

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

PR Bot seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[cubic-dev-ai]

1 issue found across 10 files

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="packages/openai-adapters/src/index.ts">

<violation number="1" location="packages/openai-adapters/src/index.ts:145">
P1: MiniMax is wired to generic `OpenAIApi`, which skips the repo’s MiniMax-specific request fixes (temperature clamping and `response_format` removal), creating a real incompatibility path in adapter-based runtime flows.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

[cubic-dev-ai]

P1: MiniMax is wired to generic OpenAIApi, which skips the repo’s MiniMax-specific request fixes (temperature clamping and response_format removal), creating a real incompatibility path in adapter-based runtime flows.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At packages/openai-adapters/src/index.ts, line 145:

<comment>MiniMax is wired to generic `OpenAIApi`, which skips the repo’s MiniMax-specific request fixes (temperature clamping and `response_format` removal), creating a real incompatibility path in adapter-based runtime flows.</comment>

<file context>
@@ -141,6 +141,8 @@ export function constructLlmApi(config: LLMConfig): BaseLlmApi | undefined {
     case "groq":
       return openAICompatible("https://api.groq.com/openai/v1/", config);
+    case "minimax":
+      return openAICompatible("https://api.minimax.io/v1/", config);
     case "sambanova":
       return openAICompatible("https://api.sambanova.ai/v1/", config);
</file context>