feat(models): add SCX.ai general-purpose provider - #3333
Open
bhuvan2134686 wants to merge 1 commit into
Open
Conversation
Add scx-ai-gp, a second SCX.ai provider covering the general-purpose (non-Turbo) tier, with GLM-5.2 as its first model. The existing scx-ai id is left untouched so no persisted value changes meaning and no data migration is required. It is renamed for display only, to "SCX.ai (Turbo)", and now declares soc2: 1 and iso27001: true — previously omitted, which meant orgs with requireSoc2 or requireIso27001 enabled silently skipped SCX under the fail-closed compliance policy. scx-ai-gp shares the api.scx.ai base URL, the LLM_SCX_AI_API_KEY credential, the OpenAI-compatible request and streaming paths and the SCX wordmark with the Turbo tier. It carries no modelCardBadge, so the badge-presence sort keeps Turbo first on shared model pages. The Turbo tier's tool_calls finish-reason normalization is deliberately not extended to it. GLM-5.2 is priced at $1.00/M input, $0.20/M cached input and $3.143/M output. maxOutput is 131072, taken from the endpoint's own validator rather than the published catalogue, which advertises a 128k ceiling.
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
Contributor
WalkthroughThe PR adds ChangesSCX.ai GP integration
Estimated code review effort: 2 (Simple) | ~10 minutes Sequence Diagram(s)sequenceDiagram
participant Client
participant Gateway
participant SCXAPI
Client->>Gateway: Send SCX.ai GP chat request
Gateway->>SCXAPI: POST /v1/chat/completions
SCXAPI-->>Gateway: Return response or stream
Gateway-->>Client: Return OpenAI-compatible output
Possibly related PRs
Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@apps/gateway/src/chat/tools/transform-streaming-to-openai.ts`:
- Line 1432: Remove scx-ai-gp from the Turbo finish-reason normalization case
group and add a separate scx-ai-gp branch that invokes transformOpenaiStreaming
without applying the end_turn, abort, or tool_use mappings. Add regression
coverage verifying all three finish reasons remain unchanged for the GP
provider.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro Plus
Run ID: 045fda5b-b187-4ce5-bff6-cf2e246687bd
📒 Files selected for processing (7)
apps/gateway/src/chat/tools/transform-response-to-openai.tsapps/gateway/src/chat/tools/transform-streaming-to-openai.tsapps/ui/src/components/provider-keys/provider-logo.tspackages/actions/src/get-provider-endpoint.tspackages/models/src/models/zai.tspackages/models/src/providers.tspackages/shared/src/components/provider-icons.tsx
| case "inference.net": | ||
| case "together-ai": | ||
| case "scx-ai": | ||
| case "scx-ai-gp": |
Contributor
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
Keep scx-ai-gp out of the Turbo finish-reason normalization branch.
Adding scx-ai-gp to this case group applies the mappings at Line [1469]-Line [1482]: end_turn becomes stop, abort becomes upstream_error, and tool_use becomes tool_calls. The PR objective states that this normalization must not apply to the GP provider.
Add a separate scx-ai-gp case that calls transformOpenaiStreaming without these mappings. Add regression tests for the three finish reasons.
Separate the GP provider case
+ case "scx-ai-gp": {
+ transformedData = transformOpenaiStreaming(
+ data,
+ usedModel,
+ supportsReasoning,
+ );
+ break;
+ }
+
case "mistral":
...
case "scx-ai":
- case "scx-ai-gp":📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| case "scx-ai-gp": | |
| case "scx-ai-gp": { | |
| transformedData = transformOpenaiStreaming( | |
| data, | |
| usedModel, | |
| supportsReasoning, | |
| ); | |
| break; | |
| } | |
| case "mistral": | |
| ... | |
| case "scx-ai": |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@apps/gateway/src/chat/tools/transform-streaming-to-openai.ts` at line 1432,
Remove scx-ai-gp from the Turbo finish-reason normalization case group and add a
separate scx-ai-gp branch that invokes transformOpenaiStreaming without applying
the end_turn, abort, or tool_use mappings. Add regression coverage verifying all
three finish reasons remain unchanged for the GP provider.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds
scx-ai-gp, a second SCX.ai provider for the general-purpose(non-Turbo) tier, with GLM-5.2 as its first model.
SCX serves two distinct tiers from
api.scx.ai. The existingscx-aiprovider represents the Turbo tier and carries the "Up to 4x faster"
model-card badge. That claim does not hold for the general-purpose tier,
so rather than widening
scx-aiand diluting its badge, this adds aseparate provider id.
The existing
scx-aiid is left untouched, so no persisted value changesmeaning and no data migration is required.
Changes to
scx-aidistinguishable in the UI. The id, badge, models and prices are
unchanged.
soc2: 1andiso27001: true. These were previously omitted,which — since the compliance policy is fail-closed — meant orgs with
requireSoc2orrequireIso27001enabled silently skipped SCX.New provider
scx-ai-gpShares the
api.scx.aibase URL, theLLM_SCX_AI_API_KEYcredential,the OpenAI-compatible request and streaming paths and the SCX wordmark
with the Turbo tier. It carries no
modelCardBadge, so the existingbadge-presence sort keeps Turbo first on shared model pages.
The Turbo tier's
tool_callsfinish-reason normalization is deliberatelynot extended to
scx-ai-gp— the general-purpose tier returns theOpenAI-standard
finish_reasonalready, soparse-provider-response.tsis unchanged.
Data policy: AU headquarters, no API training, no consumer training, no
prompt logging, 0-day retention, SOC 2 Type 1, ISO 27001.
GLM-5.2 mapping
maxOutputis set from the endpoint's own validator rather than thepublished catalogue: SCX rejects
max_tokensabove 131072 with"expected a value <= 131072", while its model card advertises a 128k
ceiling.
Verification
pnpm formatandpnpm buildpass locally; unit tests left to CI./e2eforTEST_MODELS="scx-ai-gp/glm-5.2"if a maintainer would like it.Summary by CodeRabbit