Add 3 new SiliconFlow models (GLM-4.7, GLM-4.6V, DeepSeek-V3.2)

[qychen2001]

Summary

Added 3 new SiliconFlow models to the provider database:

• zai-org/GLM-4.7 - Zhipu's new flagship model with 355B total parameters and 32B activated parameters. Supports reasoning and tool calling. Context: 205K tokens.
• zai-org/GLM-4.6V - Vision-language model with 335B total parameters. Supports image input, reasoning, and tool calling. Context: 131K tokens.
• deepseek-ai/DeepSeek-V3.2 - 671B parameter model featuring DeepSeek Sparse Attention (DSA) for faster long-context inference. Supports reasoning, structured output, and tool calling. Context: 164K tokens.

Changes

• Added model configurations for 3 new SiliconFlow models
• Context lengths are rounded to the nearest thousand following the existing convention (e.g., 204_800 → 205_000)
• All models use the standard TOML format with underscore-separated numbers
• Verified with bun validate

[Copilot]

Pull request overview

This PR adds three new SiliconFlow models to the provider database: GLM-4.7, GLM-4.6V, and DeepSeek-V3.2. The models are added with full configuration including capabilities, pricing, context limits, and modality support.

• Added GLM-4.7 (355B parameters, 205K context) with reasoning and tool calling support
• Added GLM-4.6V (335B parameters, 131K context) as a vision-language model with image input support
• Added DeepSeek-V3.2 (671B parameters, 164K context) with structured output capability

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

File	Description
providers/siliconflow/models/zai-org/GLM-4.7.toml	Adds configuration for Zhipu's GLM-4.7 flagship model with text input/output modalities
providers/siliconflow/models/zai-org/GLM-4.6V.toml	Adds configuration for GLM-4.6V vision-language model supporting text and image inputs
providers/siliconflow/models/deepseek-ai/DeepSeek-V3.2.toml	Adds configuration for DeepSeek-V3.2 with enhanced sparse attention and structured output support

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The family name should use dots instead of dashes to match the existing naming convention. GLM-4.5 uses "glm-4.5" and GLM-4.6 uses "glm-4.6", so this should be "glm-4.7" instead of "glm-4-7".

Suggested change

	family = "glm-4-7"
	family = "glm-4.7"

[Copilot]

The family name should use dots instead of dashes to match the existing naming convention. GLM-4.5V uses "glm-4.5v", so this should be "glm-4.6v" instead of "glm-4-6v".

Suggested change

	family = "glm-4-6v"
	family = "glm-4.6v"

[Copilot]

The array formatting is inconsistent with the project standard. Arrays should have spaces after commas. This should be ["text", "image"] instead of ["text","image"].

Suggested change

	input = ["text","image"]
	input = ["text", "image"]

[Copilot]

The family name should be "deepseek-v3" to match the existing pattern. Other DeepSeek V3 variants (DeepSeek-V3, DeepSeek-V3.1, DeepSeek-V3.1-Terminus, DeepSeek-V3.2-Exp) all use "deepseek-v3" as the family name, regardless of their minor version differences.

Suggested change

	family = "deepseek-v3-2"
	family = "deepseek-v3"