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 #15316] Ollama Claude Code context auto-compact/Timeout issue #35556

New Issue
Open
opened 2026-04-22 20:08:16 -05:00 by GiteaMirror · 3 comments
GiteaMirror commented 2026-04-22 20:08:16 -05:00
Owner
Copy Link

Originally created by @kausikp11 on GitHub (Apr 4, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/15316

What is the issue?

While we set the Context Length in Ollama based on GPU size when the model reach the context length the Claude is not doing AutoCompact or give warning of the Context length being filled fully. When I ran the /context in the interface it shows the output as Token used/Max token of model instead it should be Token used/Max token we have set in Ollama. This is when the Claudes auto compact or the real time out issue can be addressed.

Also you can try to see what is the max GPU size and try to fit the full context size for the model if possible?

Added screenshot for reference.

Image

Relevant log output

OS

Linux

GPU

Nvidia

CPU

Intel

Ollama version

0.18.2

Originally created by @kausikp11 on GitHub (Apr 4, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/15316 ### What is the issue? While we set the Context Length in Ollama based on GPU size when the model reach the context length the Claude is not doing AutoCompact or give warning of the Context length being filled fully. When I ran the `/context` in the interface it shows the output as `Token used/Max token of model` instead it should be `Token used/Max token we have set in Ollama`. This is when the Claudes auto compact or the real time out issue can be addressed. Also you can try to see what is the max GPU size and try to fit the full context size for the model if possible? Added screenshot for reference. <img width="1531" height="471" alt="Image" src="https://github.com/user-attachments/assets/9ad323e0-8fe9-4601-9332-e38bdc615ac4" /> ### Relevant log output ```shell ``` ### OS Linux ### GPU Nvidia ### CPU Intel ### Ollama version 0.18.2
GiteaMirror added the bug label 2026-04-22 20:08:16 -05:00
GiteaMirror commented 2026-04-22 20:08:18 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Apr 5, 2026):

This is a Claude Code issue, not an ollama issue. CC assumes that the model has a context size of 200k because that's the default for Anthropic models. In theory CC could probe the model to find out what context the model has been loaded with, but logs of HTTP requests don't show any attempt at discoverability. CC does support setting a larger context with /model gemma4[1m], but smaller values are rejected.

<!-- gh-comment-id:4188087713 --> @rick-github commented on GitHub (Apr 5, 2026): This is a Claude Code issue, not an ollama issue. CC assumes that the model has a context size of 200k because that's the [default](https://support.claude.com/en/articles/8606394-how-large-is-the-context-window-on-paid-claude-plans) for Anthropic models. In theory CC could probe the model to find out what context the model has been loaded with, but logs of HTTP requests don't show any attempt at discoverability. CC does support setting a larger context with `/model gemma4[1m]`, but smaller values are rejected.
GiteaMirror commented 2026-04-22 20:08:18 -05:00
Author
Owner
Copy Link

@ParthSareen commented on GitHub (Apr 13, 2026):

We currently set the context length correctly for cloud models but not local on purpose. I've given it some thought and had decided this as we do context shifting under the hood. The experience for local models would degrade significantly if you had to hit compaction every couple prompts. Caching doesn't work great either as the compaction has a different prompt. You can configure it yourself if you'd like to try it with CLAUDE_CODE_AUTO_COMPACT_WINDOW

<!-- gh-comment-id:4238658591 --> @ParthSareen commented on GitHub (Apr 13, 2026): We currently set the context length correctly for cloud models but not local on purpose. I've given it some thought and had decided this as we do context shifting under the hood. The experience for local models would degrade significantly if you had to hit compaction every couple prompts. Caching doesn't work great either as the compaction has a different prompt. You can configure it yourself if you'd like to try it with `CLAUDE_CODE_AUTO_COMPACT_WINDOW`
GiteaMirror commented 2026-04-22 20:08:19 -05:00
Author
Owner
Copy Link

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

🤖 Automated Triage & Analysis Report

Issue: #15316
Analyzed: 2026-04-18T18:22:37.800109

Analysis

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

Implementation Plan

  • Effort: medium
  • Steps:

This issue has been triaged and marked for implementation.

<!-- gh-comment-id:4274310342 --> @PureBlissAK commented on GitHub (Apr 18, 2026): <!-- ollama-issue-orchestrator:v1 issue:15316 --> ## 🤖 Automated Triage & Analysis Report **Issue**: #15316 **Analyzed**: 2026-04-18T18:22:37.800109 ### 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
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#35556
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?