feat(deepinfra): add GLM-4.7-Flash model

[Dagnan]

Summary

Add zai-org/GLM-4.7-Flash model configuration to the DeepInfra provider. GLM-4.7-Flash is a 30B-A3B MoE model released by Z.AI (Zhipu AI) in January 2026, positioned as the "free-ish" variant of the GLM-4.7 family with strong performance across coding, reasoning, and generative tasks.

Changes

• New file: providers/deepinfra/models/zai-org/GLM-4.7-Flash.toml
• Model ID: zai-org/GLM-4.7-Flash
• Family: glm-flash
• Pricing: $0.06 input / $0.40 output per 1M tokens
• Context window: 202,752 tokens
• Output limit: 16,384 tokens (DeepInfra enforced limit)
• Features: reasoning, tool calling, temperature control
• Open weights: true

References

• DeepInfra model page: https://deepinfra.com/zai-org/GLM-4.7-Flash
• Z.AI documentation: https://docs.z.ai/guides/llm/glm-4.7#glm-4-7-flash
• Z.AI release notes: https://docs.z.ai/release-notes/new-released (2025-01-19 - corrected to 2026-01-19)
• Z.AI blog post: https://z.ai/blog/glm-4.7

Note: The release date uses 2026-01-19 as this is the logical correct date (Z.AI's website shows a typo with 2025, but GLM-4.7 was released in December 2025, making January 2026 the correct timeline for the Flash variant).