Fixed usage handling in OpenAI.ts

[MaslovKK]

The presence of usage doesn't always mean it's the last chunk.

---

Summary by cubic

Remove special handling of usage in streaming: yield every chunk as it arrives instead of deferring usage to the end. This avoids misordering when usage appears mid-stream.

^{Written for commit c2ded53. Summary will update automatically on new commits.}

[continue]

Learn more

---

All Green is an AI agent that automatically:

✅ Addresses code review comments

✅ Fixes failing CI checks

✅ Resolves merge conflicts

[continue-development-app]

Learn more

---

All Green is an AI agent that automatically:

✅ Addresses code review comments

✅ Fixes failing CI checks

✅ Resolves merge conflicts

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[cubic-dev-ai]

1 issue found across 1 file

Prompt for AI agents (all 1 issues)

Check if these issues are valid — if so, understand the root cause of each and fix them.


<file name="packages/openai-adapters/src/apis/OpenAI.ts">

<violation number="1" location="packages/openai-adapters/src/apis/OpenAI.ts:277">
P1: `finish_reason` is not a direct property on `ChatCompletionChunk` - it&#39;s on each choice in `choices[0].finish_reason`. This condition will always be false, causing usage data to never be captured.</violation>
</file>

---

Since this is your first cubic review, here's how it works:

• cubic automatically reviews your code and comments on bugs and improvements
• Teach cubic by replying to its comments. cubic learns from your replies and gets better over time
• Ask questions if you need clarification on any suggestion

_{Reply to cubic to teach it or ask questions. Re-run a review with @cubic-dev-ai review this PR}

[cubic-dev-ai]

P1: finish_reason is not a direct property on ChatCompletionChunk - it's on each choice in choices[0].finish_reason. This condition will always be false, causing usage data to never be captured.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At packages/openai-adapters/src/apis/OpenAI.ts, line 277:

<comment>`finish_reason` is not a direct property on `ChatCompletionChunk` - it&#39;s on each choice in `choices[0].finish_reason`. This condition will always be false, causing usage data to never be captured.</comment>

<file context>
@@ -274,7 +274,7 @@ export class OpenAIApi implements BaseLlmApi {
     for await (const result of response) {
       // Check if this chunk contains usage information
-      if (result.usage) {
+      if (result.usage &amp;&amp; result.finish_reason === &quot;stop&quot;) {
         // Store it to emit after all content chunks
         lastChunkWithUsage = result;
</file context>

Suggested change

	if (result.usage && result.finish_reason === "stop") {
	if (result.usage && result.choices?.[0]?.finish_reason === "stop") {

✅ Addressed in c2ded53

[continue]

Bug Fix Required

cubic correctly identified an issue: finish_reason is not a direct property on ChatCompletionChunk. It's located at choices[0].finish_reason.

The current code:

if (result.usage && result.finish_reason === "stop")

Should be:

if (result.usage && result.choices?.[0]?.finish_reason === "stop")

I've created a branch with the fix: patch-1...continuedev:continue:fix-usage-finish-reason-access

You can either:

1. Cherry-pick the commit from that branch into your PR
2. Make the change directly in your PR

The fix aligns with how other parts of the codebase access chunk properties (e.g., line 389: chunk.choices && chunk.choices.length > 0).

[MaslovKK]

I have read the CLA Document and I hereby sign the CLA