github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-08 00:51:34 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #12780] Missing tools in new model created from qwen3-coder:30b #8477

New Issue
Closed
opened 2026-04-12 21:10:06 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-04-12 21:10:06 -05:00
Owner
Copy Link

Originally created by @somera on GitHub (Oct 25, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/12780

What is the issue?

Hi,

I created this

$ more Modelfile-qwen3-coder-30b-32k
FROM qwen3-coder:30b
PARAMETER num_ctx 32768

modelfile.

And I'm wondertind, that the new model has no tools.

Original:

$ ollama show qwen3-coder:30b
  Model
    architecture        qwen3moe
    parameters          30.5B
    context length      262144
    embedding length    2048
    quantization        Q4_K_M

  Capabilities
    completion
    tools

  Parameters
    repeat_penalty    1.05
    stop              "<|im_start|>"
    stop              "<|im_end|>"
    stop              "<|endoftext|>"
    temperature       0.7
    top_k             20
    top_p             0.8

  License
    Apache License
    Version 2.0, January 2004
    ...

New model:

$ ollama show qwen3-coder-30b-32k:latest
  Model
    architecture        qwen3moe
    parameters          30.5B
    context length      262144
    embedding length    2048
    quantization        Q4_K_M

  Capabilities
    completion

  Parameters
    temperature       0.7
    top_k             20
    top_p             0.8
    num_ctx           32768
    repeat_penalty    1.05
    stop              "<|im_start|>"
    stop              "<|im_end|>"
    stop              "<|endoftext|>"

  License
    Apache License
    Version 2.0, January 2004
    ...

Should the new model has Tools to? Cause I'm overriding only num_ctx?

And is this, what I did same with the ollama start parameter OLLAMA_CONTEXT_LENGTH=32768?

I know OLLAMA_CONTEXT_LENGTH=32768 will be activated for each model and each request.

Relevant log output

OS

Linux

GPU

Nvidia

CPU

AMD

Ollama version

0.12.6

Originally created by @somera on GitHub (Oct 25, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/12780 ### What is the issue? Hi, I created this ``` $ more Modelfile-qwen3-coder-30b-32k FROM qwen3-coder:30b PARAMETER num_ctx 32768 ``` modelfile. And I'm wondertind, that the new model has no tools. Original: ``` $ ollama show qwen3-coder:30b Model architecture qwen3moe parameters 30.5B context length 262144 embedding length 2048 quantization Q4_K_M Capabilities completion tools Parameters repeat_penalty 1.05 stop "<|im_start|>" stop "<|im_end|>" stop "<|endoftext|>" temperature 0.7 top_k 20 top_p 0.8 License Apache License Version 2.0, January 2004 ... ``` New model: ``` $ ollama show qwen3-coder-30b-32k:latest Model architecture qwen3moe parameters 30.5B context length 262144 embedding length 2048 quantization Q4_K_M Capabilities completion Parameters temperature 0.7 top_k 20 top_p 0.8 num_ctx 32768 repeat_penalty 1.05 stop "<|im_start|>" stop "<|im_end|>" stop "<|endoftext|>" License Apache License Version 2.0, January 2004 ... ``` Should the new model has Tools to? Cause I'm overriding only `num_ctx`? And is this, what I did same with the ollama start parameter `OLLAMA_CONTEXT_LENGTH=32768`? I know `OLLAMA_CONTEXT_LENGTH=32768` will be activated for each model and each request. ### Relevant log output ```shell ``` ### OS Linux ### GPU Nvidia ### CPU AMD ### Ollama version 0.12.6
GiteaMirror added the bug label 2026-04-12 21:10:06 -05:00
GiteaMirror commented 2026-04-12 21:10:08 -05:00
Author
Owner
Copy Link

@somera commented on GitHub (Oct 25, 2025):

ok, with this

FROM qwen3-coder:30b

# keep Qwen3-Coder’s chat & tool formatting
RENDERER qwen3-coder
PARSER   qwen3-coder

# your tweak
PARAMETER num_ctx 32768

Modelfile the tools are "back".

$ ollama show qwen3-coder-30b-32k:latest
  Model
    architecture        qwen3moe
    parameters          30.5B
    context length      262144
    embedding length    2048
    quantization        Q4_K_M

  Capabilities
    completion
    tools

  Parameters
    num_ctx           32768
    repeat_penalty    1.05
    stop              "<|im_start|>"
    stop              "<|im_end|>"
    stop              "<|endoftext|>"
    temperature       0.7
    top_k             20
    top_p             0.8

  License
    Apache License
    Version 2.0, January 2004
    ...

But I have open question.

And is this, what I did same with the ollama start parameter OLLAMA_CONTEXT_LENGTH=32768?
I know OLLAMA_CONTEXT_LENGTH=32768 will be activated for each model and each request.

<!-- gh-comment-id:3446838065 --> @somera commented on GitHub (Oct 25, 2025): ok, with this ``` FROM qwen3-coder:30b # keep Qwen3-Coder’s chat & tool formatting RENDERER qwen3-coder PARSER qwen3-coder # your tweak PARAMETER num_ctx 32768 ``` Modelfile the `tools` are "back". ``` $ ollama show qwen3-coder-30b-32k:latest Model architecture qwen3moe parameters 30.5B context length 262144 embedding length 2048 quantization Q4_K_M Capabilities completion tools Parameters num_ctx 32768 repeat_penalty 1.05 stop "<|im_start|>" stop "<|im_end|>" stop "<|endoftext|>" temperature 0.7 top_k 20 top_p 0.8 License Apache License Version 2.0, January 2004 ... ``` But I have open question. And is this, what I did same with the ollama start parameter `OLLAMA_CONTEXT_LENGTH=32768`? I know `OLLAMA_CONTEXT_LENGTH=32768` will be activated for each model and each request.
GiteaMirror commented 2026-04-12 21:10:08 -05:00
Author
Owner
Copy Link

@somera commented on GitHub (Oct 25, 2025):

I'm testing opencode (https://opencode.ai/) with ollama and qwen3-coder:30b. opencode is working fine wih ollama (OpenAI API) and the model (cause tools are needed). But I try to understand the difference.

When I start my ollama with OLLAMA_CONTEXT_LENGTH=32768 (or higher) it works fine.

But I can't run ollama with this context whole time. I created a new model:

FROM qwen3-coder:30b

# keep Qwen3-Coder’s chat & tool formatting
RENDERER qwen3-coder
PARSER   qwen3-coder

# your tweak
PARAMETER num_ctx 32768

But than opencode is not working. The tools want work.

What is the difference here? Work's num_ctx=32768 only when the client is using the ollama api? Cause opencode uses the compatible OpenAI api.

How can I solve it?

<!-- gh-comment-id:3447230826 --> @somera commented on GitHub (Oct 25, 2025): I'm testing opencode (https://opencode.ai/) with ollama and `qwen3-coder:30b`. opencode is working fine wih ollama (OpenAI API) and the model (cause tools are needed). But I try to understand the difference. When I start my ollama with `OLLAMA_CONTEXT_LENGTH=32768` (or higher) it works fine. But I can't run ollama with this context whole time. I created a new model: ``` FROM qwen3-coder:30b # keep Qwen3-Coder’s chat & tool formatting RENDERER qwen3-coder PARSER qwen3-coder # your tweak PARAMETER num_ctx 32768 ``` But than opencode is not working. The tools want work. What is the difference here? Work's `num_ctx=32768` only when the client is using the ollama api? Cause opencode uses the compatible OpenAI api. How can I solve it?
GiteaMirror commented 2026-04-12 21:10:09 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 26, 2025):

PARAMETER num_ctx 32768 should work for either API endpoint. You can check the size of the context of the loaded model by running ollama ps.

<!-- gh-comment-id:3447890530 --> @rick-github commented on GitHub (Oct 26, 2025): `PARAMETER num_ctx 32768` should work for either API endpoint. You can check the size of the context of the loaded model by running `ollama ps`.
GiteaMirror commented 2026-04-12 21:10:09 -05:00
Author
Owner
Copy Link

@somera commented on GitHub (Oct 26, 2025):

With ollama ps I see the size. I need to check the whole setup.

<!-- gh-comment-id:3448118873 --> @somera commented on GitHub (Oct 26, 2025): With `ollama ps` I see the size. I need to check the whole setup.
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#8477
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?