github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 16:40:08 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #15415] Total bug #35615

New Issue
Open
opened 2026-04-22 20:15:28 -05:00 by GiteaMirror · 9 comments
GiteaMirror commented 2026-04-22 20:15:28 -05:00
Owner
Copy Link

Originally created by @DjceUo on GitHub (Apr 8, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15415

What is the issue?

Ollama on previous versions worked smoothly with models like GPT-OSS-20B and similar ones such as QWEN3-30B-3A with practically no CPU load, but now they cause CPU utilization near 100% through OpenWeb UI or ChatBox with context windows of 32 or 64k tokens. Restore proper Ollama functionality.

Relevant log output

OS

Windows

GPU

Nvidia

CPU

Intel

Ollama version

0.20.3

Originally created by @DjceUo on GitHub (Apr 8, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15415 ### What is the issue? Ollama on previous versions worked smoothly with models like GPT-OSS-20B and similar ones such as QWEN3-30B-3A with practically no CPU load, but now they cause CPU utilization near 100% through OpenWeb UI or ChatBox with context windows of 32 or 64k tokens. Restore proper Ollama functionality. ### Relevant log output ```shell ``` ### OS Windows ### GPU Nvidia ### CPU Intel ### Ollama version 0.20.3
GiteaMirror added the needs more infobug labels 2026-04-22 20:15:28 -05:00
GiteaMirror commented 2026-04-22 20:15:29 -05:00
Author
Owner
Copy Link

@DjceUo commented on GitHub (Apr 8, 2026):

A compelling request to the developers: please test thoroughly before releasing Ollama. I know several sizable companies considering Ollama for local AI, but they're held back by release instability and unpredictability. Models that worked on a previous release can easily stop working.

<!-- gh-comment-id:4204253266 --> @DjceUo commented on GitHub (Apr 8, 2026): A compelling request to the developers: please test thoroughly before releasing Ollama. I know several sizable companies considering Ollama for local AI, but they're held back by release instability and unpredictability. Models that worked on a previous release can easily stop working.
GiteaMirror commented 2026-04-22 20:15:30 -05:00
Author
Owner
Copy Link

@grepin commented on GitHub (Apr 8, 2026):

@DjceUo you are tremendous! "just forget, write more tests by yourserf, or buy commercial product". it's opensource, "as is". try to understand, "feel" the ideology and keep your emotions

<!-- gh-comment-id:4204310198 --> @grepin commented on GitHub (Apr 8, 2026): @DjceUo you are tremendous! "just forget, write more tests by yourserf, or buy commercial product". it's opensource, "as is". try to understand, "feel" the ideology and keep your emotions
GiteaMirror commented 2026-04-22 20:15:30 -05:00
Author
Owner
Copy Link

@DjceUo commented on GitHub (Apr 8, 2026):

The flagship open-source Linux distro runs stably and predictably from release to release, excluding the crap builds.

<!-- gh-comment-id:4205390709 --> @DjceUo commented on GitHub (Apr 8, 2026): The flagship open-source Linux distro runs stably and predictably from release to release, excluding the crap builds.
GiteaMirror commented 2026-04-22 20:15:30 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Apr 8, 2026):

Server logs will aid in debugging.

<!-- gh-comment-id:4205675513 --> @rick-github commented on GitHub (Apr 8, 2026): [Server logs](https://docs.ollama.com/troubleshooting) will aid in debugging.
GiteaMirror commented 2026-04-22 20:15:31 -05:00
Author
Owner
Copy Link

@DjceUo commented on GitHub (Apr 9, 2026):

I can just picture Linus Torvalds saying, 'If you don't provide memory dumps, I won't fix any bugs...' Are you seriously suggesting I spend a couple of days gathering server logs for you?

<!-- gh-comment-id:4211720588 --> @DjceUo commented on GitHub (Apr 9, 2026): I can just picture Linus Torvalds saying, 'If you don't provide memory dumps, I won't fix any bugs...' Are you seriously suggesting I spend a couple of days gathering server logs for you?
GiteaMirror commented 2026-04-22 20:15:32 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Apr 9, 2026):

It will make it easier to resolve the issue. For example, the logs you provided in #14636 showed why the model load failed, and in #14849 we found that you had insufficient free VRAM. Since the issue you are having seems to be with external clients, logs will provide the basis for determining where the cause of the problem is. In the process of debugging the issue, logs from those clients may also be required.

<!-- gh-comment-id:4212503555 --> @rick-github commented on GitHub (Apr 9, 2026): It will make it easier to resolve the issue. For example, the logs you provided in #14636 showed why the model load failed, and in #14849 we found that you had insufficient free VRAM. Since the issue you are having seems to be with external clients, logs will provide the basis for determining where the cause of the problem is. In the process of debugging the issue, logs from those clients may also be required.
GiteaMirror commented 2026-04-22 20:15:32 -05:00
Author
Owner
Copy Link

@DjceUo commented on GitHub (Apr 10, 2026):

I did a quick investigation and found the following:
When using Ollama via the command line or its native GUI, everything works almost flawlessly—even when VRAM is insufficient for the context window. In such cases, Ollama seamlessly offloads to system RAM.
However, when trying to run models that require more context than fits in VRAM (thus needing RAM offloading) through ChatBox or OpenWebUI, Ollama essentially fails in these scenarios. This issue wasn't present in earlier versions. I haven't dug into the code yet.

<!-- gh-comment-id:4221945659 --> @DjceUo commented on GitHub (Apr 10, 2026): I did a quick investigation and found the following: When using Ollama via the command line or its native GUI, everything works almost flawlessly—even when VRAM is insufficient for the context window. In such cases, Ollama seamlessly offloads to system RAM. However, when trying to run models that require more context than fits in VRAM (thus needing RAM offloading) through ChatBox or OpenWebUI, Ollama essentially fails in these scenarios. This issue wasn't present in earlier versions. I haven't dug into the code yet.
GiteaMirror commented 2026-04-22 20:15:33 -05:00
Author
Owner
Copy Link

@DjceUo commented on GitHub (Apr 10, 2026):

I can't say for sure why the new models aren't working or are only partially functional. If I feel like it, I'll dig into the code over the weekend.

<!-- gh-comment-id:4221955658 --> @DjceUo commented on GitHub (Apr 10, 2026): I can't say for sure why the new models aren't working or are only partially functional. If I feel like it, I'll dig into the code over the weekend.
GiteaMirror commented 2026-04-22 20:15:33 -05:00
Author
Owner
Copy Link

@PureBlissAK commented on GitHub (Apr 18, 2026):

🤖 Automated Triage & Analysis Report

Issue: #15415
Analyzed: 2026-04-18T18:21:03.027411

Analysis

  • Type: unknown
  • Severity: medium
  • Components: unknown

Implementation Plan

  • Effort: medium
  • Steps:

This issue has been triaged and marked for implementation.

<!-- gh-comment-id:4274307633 --> @PureBlissAK commented on GitHub (Apr 18, 2026): <!-- ollama-issue-orchestrator:v1 issue:15415 --> ## 🤖 Automated Triage & Analysis Report **Issue**: #15415 **Analyzed**: 2026-04-18T18:21:03.027411 ### Analysis - **Type**: unknown - **Severity**: medium - **Components**: unknown ### Implementation Plan - **Effort**: medium - **Steps**: *This issue has been triaged and marked for implementation.*
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug needs more info
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#35615
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?