Add iFlow provider with 17 models

[shariqriazz]

Summary

• Add iFlow (iflowcn) provider with OpenAI-compatible API
• 17 models across multiple AI families with free access
• Chinese AI platform offering state-of-the-art open-source models

Provider Details

• API Base: https://apis.iflow.cn/v1
• Documentation: https://platform.iflow.cn/en/docs
• Compatibility: 100% OpenAI-compatible interface
• Pricing: Free (0.0 input/output for all models)

Models Added

Taobao Star Series

• TStars-2.0: E-commerce optimized LLM (128K context, 64K output)

Qwen3 Series

• Qwen3-Coder-Plus: Agentic coding model (256K context, 64K output)
• Qwen3-Coder-480B-A35B: Code generation specialist (256K context, 64K output)
• Qwen3-Max: General purpose flagship (256K context, 32K output)
• Qwen3-VL-Plus: Vision-language model (256K context, 32K output)
• Qwen3-Max-Preview: Preview version with enhanced capabilities (256K context, 32K output)
• Qwen3-32B: Compact general model (128K context, 32K output)
• Qwen3-235B-A22B-Thinking: Reasoning model with thinking mode (256K context, 64K output)
• Qwen3-235B-A22B-Instruct: Instruction-following variant (256K context, 64K output)
• Qwen3-235B-A22B: Dual-mode model (128K context, 32K output)

Kimi Series

• Kimi-K2: MoE model with 1T parameters (128K context, 64K output)
• Kimi-K2-Instruct-0905: Latest version (256K context, 64K output)

DeepSeek Series

• DeepSeek-V3.2-Exp: Experimental with Sparse Attention (128K context, 64K output)
• DeepSeek-V3.1-Terminus: Hybrid thinking model (128K context, 64K output)
• DeepSeek-R1: RL-trained reasoning model (128K context, 32K output)
• DeepSeek-V3-671B: MoE flagship (128K context, 32K output)

GLM Series

• GLM-4.6: Agent-optimized foundation model (200K context, 128K output)

Key Features

• Long context support (128K-256K tokens)
• Large output capabilities (32K-128K tokens)
• Reasoning models with thinking modes
• Vision-language capabilities
• Tool calling and agent support
• E-commerce and code generation specializations

Based on: https://platform.iflow.cn/en/models

[rekram1-node]

/oc update the iflow logo.svg to be consistent with other logo.svgs for providers like v0, openai, azure, etc.

The current logo.svg has custom colors that we do not want. Use fill="currentColor" for the path

[opencode-agent]

Perfect! I've successfully updated the iFlow logo.svg to be consistent with other provider logos. Here's what I changed:

Changes Made

1. Consistent structure: Updated from complex nested <g> elements with filters and gradients to a simple <svg> + <path> structure
2. Standard viewBox: Changed from viewBox="0 0 32 32" to viewBox="0 0 24 24" to match other providers
3. Removed custom colors: Eliminated the gradient definition with custom colors (#5C5CFF, #AE5CFF, #F0F2F5)
4. Added currentColor: Now uses fill="currentColor" which allows the logo to inherit the text color from its parent element
5. Simplified attributes: Removed unnecessary attributes like xmlns:xlink, fill="none", version, width, and height
6. Clean path: Converted the complex path data to work within the 24x24 viewBox while maintaining the logo's visual identity

The updated logo now follows the same pattern as other provider logos like v0, OpenAI, and Azure:

• Simple <svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24">
• Single <path fill="currentColor" d="..."/>
• Clean, consistent structure for theming support

This ensures the iFlow logo will properly inherit theme colors and display consistently with all other provider logos in the models.dev interface.

opencode session  |  github run