I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
I am using the latest version of Open WebUI.
Installation Method
Docker
Open WebUI Version
v0.8.12
Ollama Version (if applicable)
N/A
Operating System
Linux
Browser (if applicable)
Firefox
Confirmation
I have read and followed all instructions in README.md.
I am using the latest version of both Open WebUI and Ollama.
I have included the browser console logs.
I have included the Docker container logs.
I have provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.
Expected Behavior
Context remains the same
Actual Behavior
Context goes down by 10k
Steps to Reproduce
Create a model under workshop.
Use a base model.
Put 80k context into the system prompt itself.
Start a chat with the model. View the context under “usage”
Context should show 80k-82k at this point.
Switch the BASE MODEL on the workshop model.
Go back to the chat. Reply again.
Context drops by 10k-12k on the next reply.
Logs & Screenshots
N/A
Additional Information
No response
Originally created by @frenzybiscuit on GitHub (Apr 4, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/23404
### Check Existing Issues
- [x] I have searched for any existing and/or related issues.
- [x] I have searched for any existing and/or related discussions.
- [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
- [x] I am using the latest version of Open WebUI.
### Installation Method
Docker
### Open WebUI Version
v0.8.12
### Ollama Version (if applicable)
N/A
### Operating System
Linux
### Browser (if applicable)
Firefox
### Confirmation
- [x] I have read and followed all instructions in `README.md`.
- [x] I am using the latest version of **both** Open WebUI and Ollama.
- [x] I have included the browser console logs.
- [x] I have included the Docker container logs.
- [x] I have **provided every relevant configuration, setting, and environment variable used in my setup.**
- [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
- [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps:
- Start with the initial platform/version/OS and dependencies used,
- Specify exact install/launch/configure commands,
- List URLs visited, user input (incl. example values/emails/passwords if needed),
- Describe all options and toggles enabled or changed,
- Include any files or environmental changes,
- Identify the expected and actual result at each stage,
- Ensure any reasonably skilled user can follow and hit the same issue.
### Expected Behavior
Context remains the same
### Actual Behavior
Context goes down by 10k
### Steps to Reproduce
Create a model under workshop.
Use a base model.
Put 80k context into the system prompt itself.
Start a chat with the model. View the context under “usage”
Context should show 80k-82k at this point.
Switch the BASE MODEL on the workshop model.
Go back to the chat. Reply again.
Context drops by 10k-12k on the next reply.
### Logs & Screenshots
N/A
### Additional Information
_No response_
GiteaMirror
added the bug label 2026-05-05 23:36:47 -05:00
80k tokens at OpenAI is not the same as with Anthropic or Google for example
<!-- gh-comment-id:4204758964 -->
@Classic298 commented on GitHub (Apr 8, 2026):
how can i reproduce?
Different models count tokens differently
80k tokens at OpenAI is not the same as with Anthropic or Google for example
@frenzybiscuit commented on GitHub (Apr 11, 2026):
how can i reproduce?
Different models count tokens differently
80k tokens at OpenAI is not the same as with Anthropic or Google for example
I'm measuring this with llamacpp, so the token count should be the same.
Unless the actual GGUF model counts them differently?
<!-- gh-comment-id:4227773911 -->
@frenzybiscuit commented on GitHub (Apr 11, 2026):
> how can i reproduce?
>
> Different models count tokens differently
>
> 80k tokens at OpenAI is not the same as with Anthropic or Google for example
I'm measuring this with llamacpp, so the token count should be the same.
Unless the actual GGUF model counts them differently?
<!-- gh-comment-id:4227786251 -->
@frenzybiscuit commented on GitHub (Apr 11, 2026):
Anyway, to test just add this to a system prompt and then run a few conversations and change models.
[test.txt](https://github.com/user-attachments/files/26642511/test.txt)
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @frenzybiscuit on GitHub (Apr 4, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/23404
Check Existing Issues
Installation Method
Docker
Open WebUI Version
v0.8.12
Ollama Version (if applicable)
N/A
Operating System
Linux
Browser (if applicable)
Firefox
Confirmation
README.md.Expected Behavior
Context remains the same
Actual Behavior
Context goes down by 10k
Steps to Reproduce
Create a model under workshop.
Use a base model.
Put 80k context into the system prompt itself.
Start a chat with the model. View the context under “usage”
Context should show 80k-82k at this point.
Switch the BASE MODEL on the workshop model.
Go back to the chat. Reply again.
Context drops by 10k-12k on the next reply.
Logs & Screenshots
N/A
Additional Information
No response
@Classic298 commented on GitHub (Apr 8, 2026):
how can i reproduce?
Different models count tokens differently
80k tokens at OpenAI is not the same as with Anthropic or Google for example
@frenzybiscuit commented on GitHub (Apr 11, 2026):
I'm measuring this with llamacpp, so the token count should be the same.
Unless the actual GGUF model counts them differently?
@frenzybiscuit commented on GitHub (Apr 11, 2026):
Anyway, to test just add this to a system prompt and then run a few conversations and change models.
test.txt