Skip to content

feat: add Pearl as inference provider - #1813

Draft
evgeniy-pearl wants to merge 1 commit into
genlayerlabs:v0.123-devfrom
evgeniy-pearl:feat/pearl-provider
Draft

evgeniy-pearl wants to merge 1 commit into
genlayerlabs:v0.123-devfrom
evgeniy-pearl:feat/pearl-provider

Conversation

@evgeniy-pearl

Copy link
Copy Markdown

What

  • adds Pearl as a default openai-compatible provider
  • adds 4 Pearl model presets under backend/node/create_nodes/default_providers/:
    • deepseek-ai/DeepSeek-V4.1-Flash-1017
    • deepseek-ai/DeepSeek-V4-Pro-0813
    • zai-org/GLM-5.3
    • zai-org/GLM-5.3-Flash
  • adds pearl to the backend and frontend provider schemas
  • adds PEARL_API_KEY to .env.example
  • adds a unit test that keeps the pearl provider rule aligned between the backend and frontend schemas

Why

  • adding Pearl as a first-class inference provider for GenLayer validators
  • lets Studio users configure and use Pearl-hosted models out of the box

Testing done

  • verified all 4 model IDs are present on https://inference.pearlresearch.ai/v1/models
  • tested each model with a live API key at https://inference.pearlresearch.ai/v1/chat/completions, with the request body GenVM's openai-compatible provider sends:
    • plain text request: HTTP 200, finish_reason: stop, for all 4 models
    • JSON mode (response_format: json_object): HTTP 200, valid JSON, for all 4 models
  • validated the 4 preset files against both provider schemas
  • providersSchemaSync.test.ts passes, and Prettier is clean

Decisions made

  • used openai-compatible for the Pearl provider integration
  • api_url is the bare host https://inference.pearlresearch.ai, without /v1, because GenVM appends /v1/chat/completions
  • no model enum in the schema rule, which matches the arbitrary-model-name change in fix: allow arbitrary model names for all providers #1559

Checks

  • I have tested this code
  • I have reviewed my own PR
  • I have created an issue for this PR
  • I have set a descriptive PR title compliant with conventional commits

Reviewing tips

The change follows #1516 (Chutes.ai). The preset prefill in ProviderModal.vue from that PR already covers the new presets, so no component change is needed.

User facing release notes

Pearl is now available as an inference provider in Studio, with 4 models. Set PEARL_API_KEY to use it.

Add Pearl as a default openai-compatible provider with presets for
deepseek-ai/DeepSeek-V4.1-Flash-1017, deepseek-ai/DeepSeek-V4-Pro-0813,
zai-org/GLM-5.3 and zai-org/GLM-5.3-Flash, add the pearl rule to the
backend and frontend provider schemas, document PEARL_API_KEY in
.env.example, and cover the schema rule in the sync test.

api_url is the bare host because GenVM appends /v1/chat/completions.
@coderabbitai

coderabbitai Bot commented Nov 8, 2026

Copy link
Copy Markdown
Contributor

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

  • Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Tip

🎯 Customize your risk assessments

Guide how CodeRabbit assesses merge risk and security architecture risk. Highlight sensitive changes and explain your security boundaries using Merge Risk Instructions and Security Risk Instructions in your review settings. Security architecture risk assessments require the Advanced plan or above.

✅ Automatic approvals, with your rules

Let CodeRabbit approve pull requests once required threads are resolved, pre-merge checks pass, and your risk limits are met. Enable Automatic approvals in review settings, with overrides for teams or file paths.

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

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant