refactor: drop dead max_output_tokens, simplify maxTokens to 32K + pi cap

Qoder's /model/list catalog never returns max_output_tokens (the field does
not exist on any entry — only max_input_tokens is present), so the per-model
lookup `entry.max_output_tokens || 32768` was dead code that always fell back
to the 32768 literal.

- models.ts: remove max_output_tokens from QoderModelEntry; maxTokens is now a
  flat 32768 (matches qodercli's 32e3 fallback).
- stream.ts: remove the now-unused maxOutputTokens variable and its always-true
  branch; maxTokens = 32768, capped lower only by options.maxTokens (e.g. pi
  compaction passes 40000). pi-ai's maxTokens setting still has an outlet.
- stream.ts fallback modelConfig: drop the dead max_output_tokens: 32768 field.

qodercli does conditionally send max_tokens (user setting → generation config
→ 32K fallback), but the field is not required by the upstream API. Keeping it
gives pi-ai's options.maxTokens a place to flow through.
This commit is contained in:
Loongphy
2026-08-28 19:07:16 +08:00
parent 5393c24b7e
commit 112da4deaf
2 changed files with 4 additions and 7 deletions
+1 -2
View File
@@ -19,7 +19,6 @@ export interface QoderModelEntry {
enable?: boolean; enable?: boolean;
display_name?: string; display_name?: string;
max_input_tokens?: number; max_input_tokens?: number;
max_output_tokens?: number;
context_config?: Record<string, { token_count?: number; is_default?: boolean }>; context_config?: Record<string, { token_count?: number; is_default?: boolean }>;
is_vl?: boolean; is_vl?: boolean;
is_reasoning?: boolean; is_reasoning?: boolean;
@@ -589,7 +588,7 @@ export async function updateQoderModelsCache(
input: isVL ? ["text", "image"] : ["text"], input: isVL ? ["text", "image"] : ["text"],
cost: ZERO_COST, cost: ZERO_COST,
contextWindow: ctxLen, contextWindow: ctxLen,
maxTokens: entry.max_output_tokens || 32768, maxTokens: 32768,
}); });
} }
+3 -5
View File
@@ -128,7 +128,6 @@ export function streamQoder(
const modelConfig = getCachedModelConfig(model.id, providerMode) || { const modelConfig = getCachedModelConfig(model.id, providerMode) || {
key: model.id, key: model.id,
is_reasoning: false, is_reasoning: false,
max_output_tokens: 32768,
source: "system", source: "system",
}; };
// Use the cached entry's original upstream key when available; fall back to // Use the cached entry's original upstream key when available; fall back to
@@ -136,7 +135,6 @@ export function streamQoder(
const qoderModel = modelConfig.key || model.id; const qoderModel = modelConfig.key || model.id;
const isReasoning = !!modelConfig.is_reasoning; const isReasoning = !!modelConfig.is_reasoning;
const maxOutputTokens = modelConfig.max_output_tokens || 32768;
const normalizedMessages = transformMessagesForQoder(context.messages); const normalizedMessages = transformMessagesForQoder(context.messages);
const systemText = context.systemPrompt || ""; const systemText = context.systemPrompt || "";
@@ -163,10 +161,10 @@ export function streamQoder(
? `${stablePart}-${options.sessionId}` ? `${stablePart}-${options.sessionId}`
: `${stablePart}-${crypto.randomUUID()}`; : `${stablePart}-${crypto.randomUUID()}`;
// Qoder's catalog exposes no per-model output cap (no max_output_tokens),
// so we default to 32K (matches qodercli's 32e3 fallback) and let pi cap it
// lower when the caller sets options.maxTokens (e.g. compaction at 40K).
let maxTokens = 32768; let maxTokens = 32768;
if (maxOutputTokens > 0) {
maxTokens = maxOutputTokens;
}
if (options?.maxTokens && options.maxTokens < maxTokens) { if (options?.maxTokens && options.maxTokens < maxTokens) {
maxTokens = options.maxTokens; maxTokens = options.maxTokens;
} }