Update deepseek-chat context window size 64k -> 128k

[NikolayXHD]

Context

Update deepseek-chat model config to reflect increased context window size of DeepSeek-V3 model.

Although the change is not yet reflected in DeepSeek API docs, where it shows 64k, deepseek-chat points at DeepSeek-V3 which has context window 128k,see https://github.com/deepseek-ai/DeepSeek-V3

Testing

I have built the extension with the change, installed .vsix and continued a previous task so that the context window exceeded past 64k tokens, and then made another request to make sure DeepSeek API does not fail on exceeding the documented but incorrect limit of 64k

---

Important

Update deepseek-chat model's contextWindow in api.ts from 64,000 to 128,000 tokens to match DeepSeek-V3 capabilities.

• Behavior:
  • Update contextWindow for deepseek-chat model in api.ts from 64,000 to 128,000 tokens.
  • Ensures compatibility with DeepSeek-V3 model's actual context window size.
• Testing:
  • Verified API handles requests exceeding 64,000 tokens without errors.

^{This description was created by for 2f4a129. It will automatically update as commits are pushed.}

[changeset-bot]

⚠️ No Changeset found

Latest commit: ccf8af5

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[samhvw8]

Due to the low usage of requests exceeding 64K tokens, this change will help optimize server resource utilization.

@NikolayXHD check this, deepseek-ai/DeepSeek-V3#186

and this too, https://openrouter.ai/deepseek/deepseek-chat-v3-0324

it's really 64k, not mistake

[NikolayXHD]

@samhvw8 there seems to be confusion there, because conversation you refer to is January, while V3 release in API must have happened on march 24.

Additionally I did test the change.

Anyways I am fine using the updated settings in my private build.

[samhvw8]

@samhvw8 there seems to be confusion there, because conversation you refer to is January, while V3 release in API must have happened on march 24.

Additionally I did test the change.

Anyways I am fine using the updated settings in my private build.

@NikolayXHD let me check it open more time

do you have discord, we can have a conversation in that

i check in deepseek discord but they talk in chinese about 128k

[NikolayXHD]

@samhvw8 turns out I do have discord from olden times :) it's NikolayHD

[NikolayXHD]

@samhvw8 I have tested API limit explicitly by Python script and indeed the limit is 64k as per current API documentation. So, will close this PR.

It's bizarre that my indirect test with long Roo-Code task did not fail as well, but this is another story for me to investigate.

[samhvw8]

@samhvw8 I have tested API limit explicitly by Python script and indeed the limit is 64k as per current API documentation. So, will close this PR.

It's bizarre that my indirect test with long Roo-Code task did not fail as well, but this is another story for me to investigate.

yeah that strange, we can investigate toghether, btw did you join Roo's Discord ?

[NikolayXHD]

btw did you join Roo's Discord ?

yes, nikolayhd8331