Skip to content

fix: track reasoning by output_index for copilot compatibility #9124

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
rekram1-node merged 3 commits into from
Jan 19, 2026
+46 −27
Merged

fix: track reasoning by output_index for copilot compatibility #9124

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
eef69d0
fix: track reasoning by output_index for copilot compatibility
NateSmyth Jan 17, 2026
cb93d50
fix: route compatible GPT-5* models to /responses
NateSmyth Jan 17, 2026
d85e1d3
Merge branch 'dev' into copilot-reasoning-ids
rekram1-node Jan 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 7 additions & 7 deletions packages/opencode/src/provider/provider.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -615,13 +615,13 @@ export namespace Provider {
},
experimentalOver200K: model.cost?.context_over_200k
? {
cache: {
read: model.cost.context_over_200k.cache_read ?? 0,
write: model.cost.context_over_200k.cache_write ?? 0,
},
input: model.cost.context_over_200k.input,
output: model.cost.context_over_200k.output,
}
cache: {
read: model.cost.context_over_200k.cache_read ?? 0,
write: model.cost.context_over_200k.cache_write ?? 0,
},
input: model.cost.context_over_200k.input,
output: model.cost.context_over_200k.output,
}
: undefined,
},
limit: {
Expand Down
59 changes: 39 additions & 20 deletions ...ncode/src/provider/sdk/openai-compatible/src/responses/openai-responses-language-model.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -815,14 +815,20 @@ export class OpenAIResponsesLanguageModel implements LanguageModelV2 {
// flag that checks if there have been client-side tool calls (not executed by openai)
let hasFunctionCall = false

// Track reasoning by output_index instead of item_id
// GitHub Copilot rotates encrypted item IDs on every event
const activeReasoning: Record<
string,
number,
{
canonicalId: string // the item.id from output_item.added
encryptedContent?: string | null
summaryParts: number[]
}
> = {}

// Track current active reasoning output_index for correlating summary events
let currentReasoningOutputIndex: number | null = null

// Track a stable text part id for the current assistant message.
// Copilot may change item_id across text deltas; normalize to one id.
let currentTextId: string | null = null
Expand Down Expand Up @@ -933,10 +939,12 @@ export class OpenAIResponsesLanguageModel implements LanguageModelV2 {
},
})
} else if (isResponseOutputItemAddedReasoningChunk(value)) {
activeReasoning[value.item.id] = {
activeReasoning[value.output_index] = {
canonicalId: value.item.id,
encryptedContent: value.item.encrypted_content,
summaryParts: [0],
}
currentReasoningOutputIndex = value.output_index

controller.enqueue({
type: "reasoning-start",
Expand Down Expand Up @@ -1091,22 +1099,25 @@ export class OpenAIResponsesLanguageModel implements LanguageModelV2 {
currentTextId = null
}
} else if (isResponseOutputItemDoneReasoningChunk(value)) {
const activeReasoningPart = activeReasoning[value.item.id]
const activeReasoningPart = activeReasoning[value.output_index]
if (activeReasoningPart) {
for (const summaryIndex of activeReasoningPart.summaryParts) {
controller.enqueue({
type: "reasoning-end",
id: `${value.item.id}:${summaryIndex}`,
id: `${activeReasoningPart.canonicalId}:${summaryIndex}`,
providerMetadata: {
openai: {
itemId: value.item.id,
itemId: activeReasoningPart.canonicalId,
reasoningEncryptedContent: value.item.encrypted_content ?? null,
},
},
})
}
delete activeReasoning[value.output_index]
if (currentReasoningOutputIndex === value.output_index) {
currentReasoningOutputIndex = null
}
}
delete activeReasoning[value.item.id]
}
} else if (isResponseFunctionCallArgumentsDeltaChunk(value)) {
const toolCall = ongoingToolCalls[value.output_index]
Expand Down Expand Up @@ -1198,32 +1209,40 @@ export class OpenAIResponsesLanguageModel implements LanguageModelV2 {
logprobs.push(value.logprobs)
}
} else if (isResponseReasoningSummaryPartAddedChunk(value)) {
const activeItem =
currentReasoningOutputIndex !== null ? activeReasoning[currentReasoningOutputIndex] : null

// the first reasoning start is pushed in isResponseOutputItemAddedReasoningChunk.
if (value.summary_index > 0) {
activeReasoning[value.item_id]?.summaryParts.push(value.summary_index)
if (activeItem && value.summary_index > 0) {
activeItem.summaryParts.push(value.summary_index)

controller.enqueue({
type: "reasoning-start",
id: `${value.item_id}:${value.summary_index}`,
id: `${activeItem.canonicalId}:${value.summary_index}`,
providerMetadata: {
openai: {
itemId: value.item_id,
reasoningEncryptedContent: activeReasoning[value.item_id]?.encryptedContent ?? null,
itemId: activeItem.canonicalId,
reasoningEncryptedContent: activeItem.encryptedContent ?? null,
},
},
})
}
} else if (isResponseReasoningSummaryTextDeltaChunk(value)) {
controller.enqueue({
type: "reasoning-delta",
id: `${value.item_id}:${value.summary_index}`,
delta: value.delta,
providerMetadata: {
openai: {
itemId: value.item_id,
const activeItem =
currentReasoningOutputIndex !== null ? activeReasoning[currentReasoningOutputIndex] : null

if (activeItem) {
controller.enqueue({
type: "reasoning-delta",
id: `${activeItem.canonicalId}:${value.summary_index}`,
delta: value.delta,
providerMetadata: {
openai: {
itemId: activeItem.canonicalId,
},
},
},
})
})
}
} else if (isResponseFinishedChunk(value)) {
finishReason = mapOpenAIResponseFinishReason({
finishReason: value.response.incomplete_details?.reason,
Expand Down