feat: Add IO Intelligence as translation provider

[ertan2002]

Context

IO Intelligence (io.intelligence) is io.net’s inference platform that lets you run multiple open-source LLMs through a single OpenAI-compatible API, backed by io.net’s decentralized GPU network. :contentReference[oaicite:0]{index=0}
It also includes a generous free daily allowance, including 500,000 tokens per model per day for API usage.
https://io.net/intelligence

Changes in this PR

• New LLM-based provider IO Intelligence using OpenAI-compatible API
• Base URL: https://api.intelligence.io.solutions/api/v1/chat/completions
• Dynamic model selection via dropdown with Refresh button
• Models fetched from /api/v1/models endpoint (no auth required)

[ertan2002]

If you want to use IO Intelligence in your projects, you can also check out my open-source NuGet package:

https://www.nuget.org/packages/IONET.IOIntelligence

IO Intelligence currently offers 16 models, with FREE 500k tokens per model per day. I’m also using it in my personal projects. In addition to the API, it also provides a web UI.

https://ai.io.net/ai/models

IO Intelligence provides 18Agents as well.
https://ai.io.net/ai/agents

[A-nony-mous]

If this API port supports the OpenAI format, why is it necessary to import other NuGet packages separately? Avoid creating entities unless absolutely necessary.

[ertan2002]

If this API port supports the OpenAI format, why is it necessary to import other NuGet packages separately? Avoid creating entities unless absolutely necessary.

I guess, there is a missunderstanding. I have not added any nuget package into my changes.

It was an extra information for people who want to use IO Intelligence for their prive projects. It was a complete different topic and just wanted to mention about it.

[A-nony-mous]

If this API port supports the OpenAI format, why is it necessary to import other NuGet packages separately? Avoid creating entities unless absolutely necessary.

I guess, there is a missunderstanding. I have not added any nuget package into my changes.

It was an extra information for people who want to use IO Intelligence for their prive projects. It was a complete different topic and just wanted to mention about it.

I still hold the same view: if this API supports the OpenAI protocol, why should I prepare a separate configuration item for it? Do I need to configure different entry points for all the different vendors such as Grok, Groq, Cerebras, Deepseek, etc.?

This is my final view. CC @SakiRinn

[ertan2002]

If this API port supports the OpenAI format, why is it necessary to import other NuGet packages separately? Avoid creating entities unless absolutely necessary.

I guess, there is a missunderstanding. I have not added any nuget package into my changes.
It was an extra information for people who want to use IO Intelligence for their prive projects. It was a complete different topic and just wanted to mention about it.

I still hold the same view: if this API supports the OpenAI protocol, why should I prepare a separate configuration item for it? Do I need to configure different entry points for all the different vendors such as Grok, Groq, Cerebras, Deepseek, etc.?

This is my final view. CC @SakiRinn

Many API providers aim to be OpenAI-compatible now. It is basically become a standard.

That said, IO Intelligence isn’t just another OpenAI-compatible endpoint. It’s powered by io.net, it operates its own GPU infrastructure, and it’s actively expanding into agents and its own models. It also supports adding, training, and running custom models. Once you get into those scenarios, we can’t always rely on one pre-defined “OpenAI-compatible” logic layer, because the protocol may be the same but the capabilities and behavior can diverge beyond it.

Also, IO Intelligence is partnering with a lot of vibe-coding extensions, and these tools generally don’t treat “OpenAI-compatible” as the only configuration path. Even when they support OpenAI-compatible providers, they still expose major providers as explicit options. That approach is pretty normal in more mature products because it keeps the UX clearer, leaves room for provider-specific settings, and makes the system more flexible as providers evolve.

Here are IO Intelligence’s partners

And here are a couple of examples showing how other tools handle it:

Cline:

Roo Code:

… and there are more like this.

[SakiRinn]

Thank you for your contribution! The code quality looks excellent. However, after careful consideration of the project's long-term plan, I agree that we should not merge this specific integration for the moment.

To ensure the maintainability of LiveCaptions-Translator, our current strategy is to restrict hard-coded integrations for specific vendors, especially those that fully comply with OpenAI API standards. By the way, we integrated OpenRouter because it serves as a model aggregator and a unified gateway. In contrast, adding individual inference platforms separately (such as IO Intelligence, and potentially Together, Anyscale, DeepInfra, etc.) would bloat the codebase and significantly increase the maintenance overhead.

Given that IO Intelligence provides OpenAI-compatible endpoints, I recommend that users access this service through the generic "Custom OpenAI" / "Custom Endpoint" feature in the settings. This way, users can still take advantage of IO Intelligence's services (and their free quota) without requiring specific adaptation in the core repository.

[A-nony-mous]

Close as not expected