smashimo/oh-my-pi
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge PR #8852: fix(catalog): add deepseek-v4-pro-0813 discovery limits (@tommyldev)

Browse Source
This commit is contained in:
can1357
2026-08-19 01:37:00 +02:00
parent 60079ffaea 42171bf16e
commit 381ed55d5b
3 changed files with 9 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
packages/catalog/CHANGELOG.md
+1
View File
@@ -5,6 +5,7 @@
### Fixed
- Fixed local Qwen 3.8+ models (llama.cpp, vLLM, loopback custom providers) exposing the generic `minimal..high` thinking ladder instead of the chat template's real `low`/`medium`/`xhigh` `reasoning_effort` tiers. The derived metadata now marks thinking as mandatory (the official 3.8 template raises on `enable_thinking: false`), vLLM-served Qwen routes through the `chat_template_kwargs` dialect (top-level `enable_thinking` is ignored by vLLM), and vLLM discovery lights up the reasoning dial for Qwen 3.8+ ids its `/v1/models` endpoint reports as non-reasoning.
- Fixed `deepseek-v4-pro-0813` surfacing from Alibaba Token Plan discovery with `contextWindow`/`maxTokens` of `null`. The dated DeepSeek V4 Pro snapshot was missing from `ALIBABA_TOKEN_PLAN_DISCOVERED_MODEL_LIMITS`, so unlike its `deepseek-v4-flash-0731` sibling it fell through to unknown limits ([#8847](https://github.com/can1357/oh-my-pi/issues/8847)).
## [17.3.6] - 2026-08-17
packages/catalog/src/provider-models/openai-compat.ts
+4
View File
@@ -2981,6 +2981,10 @@ export const ALIBABA_TOKEN_PLAN_DISCOVERED_MODEL_LIMITS: Readonly<Record<string,
contextWindow: 1_000_000,
maxTokens: 384_000,
},
"deepseek-v4-pro-0813": {
contextWindow: 1_000_000,
maxTokens: 384_000,
},
"deepseek-v3.2": {
contextWindow: 131_072,
maxTokens: 65_536,
packages/catalog/test/alibaba-token-plan.test.ts
+4 -1
View File
@@ -75,6 +75,7 @@ describe("QwenCloud Token Plan provider", () => {
},
{ id: "deepseek-v4-flash", owned_by: "qwencloud" },
{ id: "deepseek-v4-flash-0731", owned_by: "qwencloud" },
{ id: "deepseek-v4-pro-0813", owned_by: "qwencloud" },
{ id: "kimi-k2.7-code", owned_by: "qwencloud" },
{ id: "MiniMax-M2.5", owned_by: "qwencloud" },
{ id: "qwen3.6-plus", owned_by: "qwencloud" },
@@ -106,6 +107,7 @@ describe("QwenCloud Token Plan provider", () => {
"deepseek-v3.2",
"deepseek-v4-flash",
"deepseek-v4-flash-0731",
"deepseek-v4-pro-0813",
"future-chat-model",
"glm-5",
"glm-5.1",
@@ -122,6 +124,7 @@ describe("QwenCloud Token Plan provider", () => {
["qwen3.8-max", 1_000_000, 131_072],
["deepseek-v4-flash", 1_000_000, 384_000],
["deepseek-v4-flash-0731", 1_000_000, 384_000],
["deepseek-v4-pro-0813", 1_000_000, 384_000],
["deepseek-v3.2", 131_072, 65_536],
["glm-5.1", 202_752, 128_000],
["glm-5", 202_752, 16_384],
@@ -133,7 +136,7 @@ describe("QwenCloud Token Plan provider", () => {
for (const [id, contextWindow, maxTokens] of expectedLimits) {
expect(models?.find(model => model.id === id)).toMatchObject({ contextWindow, maxTokens });
}
for (const id of ["deepseek-v4-flash", "deepseek-v4-flash-0731"]) {
for (const id of ["deepseek-v4-flash", "deepseek-v4-flash-0731", "deepseek-v4-pro-0813"]) {
expect(models?.find(model => model.id === id)).toMatchObject({
reasoning: true,
thinking: {