Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-21 01:25:28 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #5534] Bug: Changing DeepSeek model or API key resets conversation context #5152

New Issue
Open
opened 2026-06-05 14:52:19 -04:00 by yindo · 1 comment
yindo commented 2026-06-05 14:52:19 -04:00
Owner
Copy Link
Copy Source

Originally created by @miraitech24 on GitHub (Apr 25, 2026).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5534

Problem Description
When I change the DeepSeek model (e.g., from deepseek-v4-flash to deepseek-v4-pro) or regenerate the API key, the workspace treats every new message as a first-time interaction. The LLM does not remember any previous conversation history, even though the old messages are still visible in the UI.

Steps to Reproduce

  1. Create a workspace
  2. Have a conversation: 'My name is Miraitech. Please remember it.' → Model responds 'Got it, Miraitech.'
  3. Ask: 'What is my name?' → Model correctly answers 'Miraitech.'
  4. Go to Settings → LLM Provider → Change model from deepseek-v4-flash to deepseek-v4-pro (or change API key)
  5. Save and return to workspace
  6. Ask again: 'Do you remember my name from our earlier chat?'
  7. Model responds 'I don't have any previous context' or similar

Expected Behavior
Changing model or API key should NOT break conversation continuity. The workspace should maintain its history and continue sending full context to the new model/key.

Actual Behavior

  • Old messages remain visible in UI
  • But they are NOT being sent to the API as context
  • Every message becomes a 'first meeting'

Cross-Validation
The same conversation flow with OpenAI GPT-3.5 (same AnythingLLM version, same workspace) works correctly — model remembers after API key changes. This suggests the issue is specific to how AnythingLLM handles the DeepSeek provider or model switching.

Technical Note
I understand LLM APIs are stateless by design. The application's job is to manage state. If AnythingLLM fails to send conversation history after settings changes, it's not fulfilling its core purpose.

Environment

  • AnythingLLM version: [please fill]
  • OS: [please fill]
  • DeepSeek API with deepseek-v4-flash / deepseek-v4-pro
  • Workspace setting: 'Use workspace LLM settings'

Screenshots/Logs
[Attach if available]

Originally created by @miraitech24 on GitHub (Apr 25, 2026). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/5534 **Problem Description** When I change the DeepSeek model (e.g., from deepseek-v4-flash to deepseek-v4-pro) or regenerate the API key, the workspace treats every new message as a first-time interaction. The LLM does not remember any previous conversation history, even though the old messages are still visible in the UI. **Steps to Reproduce** 1. Create a workspace 2. Have a conversation: 'My name is Miraitech. Please remember it.' → Model responds 'Got it, Miraitech.' 3. Ask: 'What is my name?' → Model correctly answers 'Miraitech.' 4. Go to Settings → LLM Provider → Change model from deepseek-v4-flash to deepseek-v4-pro (or change API key) 5. Save and return to workspace 6. Ask again: 'Do you remember my name from our earlier chat?' 7. Model responds 'I don't have any previous context' or similar **Expected Behavior** Changing model or API key should NOT break conversation continuity. The workspace should maintain its history and continue sending full context to the new model/key. **Actual Behavior** - Old messages remain visible in UI - But they are NOT being sent to the API as context - Every message becomes a 'first meeting' **Cross-Validation** The same conversation flow with OpenAI GPT-3.5 (same AnythingLLM version, same workspace) works correctly — model remembers after API key changes. This suggests the issue is specific to how AnythingLLM handles the DeepSeek provider or model switching. **Technical Note** I understand LLM APIs are stateless by design. The application's job is to manage state. If AnythingLLM fails to send conversation history after settings changes, it's not fulfilling its core purpose. **Environment** - AnythingLLM version: [please fill] - OS: [please fill] - DeepSeek API with deepseek-v4-flash / deepseek-v4-pro - Workspace setting: 'Use workspace LLM settings' **Screenshots/Logs** [Attach if available]
yindo added the needs info / can't replicate label 2026-06-05 14:52:19 -04:00
yindo commented 2026-06-05 14:52:20 -04:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Apr 25, 2026):

Changing model or API key should NOT break conversation continuity. The workspace should maintain its history and continue sending full context to the new model/key.

This is just not how the system even works. The API key and model are fully disconnected to how messages function. Something else is going on here and I think it is model/provider specific.

This is also the same as #5533 and I wanted to keep the conversation on a single thread until it is fully confirmed this is a bug and not a nuance we cannot control

<!-- gh-comment-id:4320485962 --> @timothycarambat commented on GitHub (Apr 25, 2026): > Changing model or API key should NOT break conversation continuity. The workspace should maintain its history and continue sending full context to the new model/key. This is just not how the system even works. The API key and model are fully disconnected to how messages function. Something else is going on here and I think it is model/provider specific. This is also the same as #5533 and I wanted to keep the conversation on a single thread until it is fully confirmed this is a bug and not a nuance we cannot control
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label needs info / can't replicate
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#5152
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?