Skip to content

fix: update DeepSeek models context window to 128k #7269

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
merged 4 commits into from
Aug 21, 2025
Merged

fix: update DeepSeek models context window to 128k #7269

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
e2cd924
fix: update DeepSeek models context window to 128k
roomote Aug 21, 2025
46ca1de
feat: update deepseek-reasoner maxTokens to 64K based on official doc…
roomote Aug 21, 2025
09903a0
fix: use default maxTokens values instead of maximum for DeepSeek models
daniel-lxs Aug 21, 2025
a57187b
fix: use maximum output tokens for both DeepSeek models
daniel-lxs Aug 21, 2025
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
10 changes: 5 additions & 5 deletions packages/types/src/providers/deepseek.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,8 +7,8 @@ export const deepSeekDefaultModelId: DeepSeekModelId = "deepseek-chat"

export const deepSeekModels = {
"deepseek-chat": {
maxTokens: 8192,
contextWindow: 64_000,
maxTokens: 8192, // 8K max output
contextWindow: 128_000,
Copy link
Contributor Author

@roomote roomote bot Aug 21, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The context window update to 128k looks correct and aligns with the DeepSeek API documentation. Consider adding a comment referencing the API docs URL (https://api-docs.deepseek.com/quick_start/pricing/) for future maintainers.

Also, this capability update might warrant a changelog entry so users know about the increased context window.

supportsImages: false,
supportsPromptCache: true,
inputPrice: 0.27, // $0.27 per million tokens (cache miss)
Expand All @@ -18,15 +18,15 @@ export const deepSeekModels = {
description: `DeepSeek-V3 achieves a significant breakthrough in inference speed over previous models. It tops the leaderboard among open-source models and rivals the most advanced closed-source models globally.`,
},
"deepseek-reasoner": {
maxTokens: 8192,
contextWindow: 64_000,
maxTokens: 65536, // 64K max output for reasoning mode
contextWindow: 128_000,
supportsImages: false,
supportsPromptCache: true,
inputPrice: 0.55, // $0.55 per million tokens (cache miss)
outputPrice: 2.19, // $2.19 per million tokens
cacheWritesPrice: 0.55, // $0.55 per million tokens (cache miss)
cacheReadsPrice: 0.14, // $0.14 per million tokens (cache hit)
description: `DeepSeek-R1 achieves performance comparable to OpenAI-o1 across math, code, and reasoning tasks. Supports Chain of Thought reasoning with up to 32K tokens.`,
description: `DeepSeek-R1 achieves performance comparable to OpenAI-o1 across math, code, and reasoning tasks. Supports Chain of Thought reasoning with up to 64K output tokens.`,
},
} as const satisfies Record<string, ModelInfo>

Expand Down
18 changes: 16 additions & 2 deletions src/api/providers/__tests__/deepseek.spec.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -154,12 +154,26 @@ describe("DeepSeekHandler", () => {
const model = handler.getModel()
expect(model.id).toBe(mockOptions.apiModelId)
expect(model.info).toBeDefined()
expect(model.info.maxTokens).toBe(8192)
expect(model.info.contextWindow).toBe(64_000)
expect(model.info.maxTokens).toBe(8192) // deepseek-chat has 8K max
expect(model.info.contextWindow).toBe(128_000)
expect(model.info.supportsImages).toBe(false)
expect(model.info.supportsPromptCache).toBe(true) // Should be true now
})

it("should return correct model info for deepseek-reasoner", () => {
const handlerWithReasoner = new DeepSeekHandler({
...mockOptions,
apiModelId: "deepseek-reasoner",
})
const model = handlerWithReasoner.getModel()
expect(model.id).toBe("deepseek-reasoner")
expect(model.info).toBeDefined()
expect(model.info.maxTokens).toBe(65536) // deepseek-reasoner has 64K max
expect(model.info.contextWindow).toBe(128_000)
expect(model.info.supportsImages).toBe(false)
expect(model.info.supportsPromptCache).toBe(true)
})

it("should return provided model ID with default model info if model does not exist", () => {
const handlerWithInvalidModel = new DeepSeekHandler({
...mockOptions,
Expand Down
Loading