feat: add ByteDance provider for GLM-4.7 and GPT-OSS-120B models

[rcogal]

Summary

Adds ByteDance (BytePlus ModelArk) as a provider for two models with verified pricing and tested configurations:

• GLM-4.7 (glm-4-7-251222)
• GPT-OSS-120B (gpt-oss-120b-250805)

Models Added

GLM-4.7

• Provider: ByteDance (BytePlus ModelArk)
• Model Name: glm-4-7-251222
• Pricing: $0.6/M input, $2.2/M output, $0.11/M cached input
• Context: 200K tokens
• Max Output: 128K tokens
• Capabilities: Streaming ✅, Reasoning ✅, Tools ✅, Caching ✅, Vision ❌, JSON output ❌

GPT-OSS-120B

• Provider: ByteDance (BytePlus ModelArk)
• Model Name: gpt-oss-120b-250805
• Pricing: $0.1/M input, $0.5/M output, $0.02/M cached input
• Context: 128K tokens
• Max Output: 32K tokens
• Capabilities: Streaming ✅, Reasoning ✅, Tools ✅, Caching ✅, Vision ❌, JSON output ❌

Notes

• Both models set jsonOutput: false because BytePlus ModelArk doesn't support the response_format parameter
• Pricing verified from BytePlus ModelArk official documentation
• Model specifications verified from BytePlus documentation

Test Plan

• GLM-4.7: All 44 E2E tests passed
  • ✅ Basic chat completions
  • ✅ Streaming responses
  • ✅ Reasoning mode
  • ✅ Tool/function calls
  • ✅ Prompt caching
• GPT-OSS-120B: All 44 E2E tests passed
  • ✅ Basic chat completions
  • ✅ Streaming responses
  • ✅ Reasoning mode
  • ✅ Tool/function calls
  • ✅ Prompt caching
• Code formatted with pnpm format
• Production build passes with pnpm build

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features
  • Added ByteDance provider support for GPT OSS 120B with a 128K context window, streaming, reasoning, and tooling capabilities.
  • Added ByteDance provider support for GLM-4.7 with a 200K context window, streaming, reasoning, and tooling capabilities.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Walkthrough

Added ByteDance provider entries to two existing model definitions: a BytePlus variant for gpt-oss-120b in openai.ts and a bytedance provider for glm-4.7 in zai.ts, including pricing, capacities, and feature flags.

Changes

Cohort / File(s)	Summary
ByteDance provider additions packages/models/src/models/openai.ts, packages/models/src/models/zai.ts	Appended new provider objects for ByteDance: gpt-oss-120b-250805 (inputPrice 0.1/1e6, outputPrice 0.5/1e6, contextSize 128000, maxOutput 32000, streaming/tools/reasoning true, jsonOutput false) and glm-4-7-251222 (inputPrice 0.6/1e6, outputPrice 2.2/1e6, contextSize 200000, maxOutput 128000, streaming/tools/reasoning true, jsonOutput false). No other logic or API changes.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• feat(models): add support for GPT OSS 120B and 20B #578: Adds the "gpt-oss-120b" model that this PR adds a ByteDance provider variant for.
• feat(models): add GLM-4.7 #1346: Adds provider entries to the GLM-4.7 model; touches the same providers array updated here.
• feat(models): add Cerebras GLM-4.7 #1415: Also modifies the GLM-4.7 providers list and may overlap with provider metadata edits.

Suggested reviewers

• smakosh

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main change: adding a ByteDance provider for two models (GLM-4.7 and GPT-OSS-120B). It is concise, clear, and directly reflects the primary objective of the PR.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}