github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #10879] codestral doesn't allow tool calling #69208

New Issue
Closed
opened 2026-05-04 17:28:37 -05:00 by GiteaMirror · 4 comments
GiteaMirror commented 2026-05-04 17:28:37 -05:00
Owner
Copy Link

Originally created by @nickkaltner on GitHub (May 27, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/10879

What is the issue?

when i use codestral, the chat endpoint works fine with model and messages parameters. When i add the tools parameter, it 400s.
I'm using the elixir ollama library, and it gives
%Ollama.HTTPError{status: 400, message: "Bad Request"}

here is the example script

model = "codestral:22b-v0.1-q2_K"

stock_price_tool = %{
  type: "function",
  function: %{
    name: "get_stock_price",
    description: "Fetches the live stock price for the given ticker.",
    parameters: %{
      type: "object",
      properties: %{
        ticker: %{
          type: "string",
          description: "The ticker symbol of a specific stock."
        }
      },
      required: ["ticker"]
    }
  }
}

messages = [
  %{role: "system", content: "You are a helpful assistant."},
  %{role: "user", content: "what is apple's latest stock price?"}
]

{:ok, resp} = Ollama.chat(client,
  model: model,
  messages: messages,
  tools: [stock_price_tool],
  keep_alive: 10000
)

IO.puts inspect(resp)

IO.puts resp["message"]["content"]

resp

If I comment out the tools: [stock_price_tool] it works without the tool use

Relevant log output

[GIN] 2025/05/27 - 21:19:14 | 400 |   20.177917ms |       127.0.0.1 | POST     "/api/chat"
[GIN] 2025/05/27 - 21:19:29 | 400 |   18.747583ms |       127.0.0.1 | POST     "/api/chat"
[GIN] 2025/05/27 - 21:26:01 | 400 |   18.512958ms |       127.0.0.1 | POST     "/api/chat"

OS

macOS

GPU

Apple

CPU

Apple

Ollama version

0.7.1

Originally created by @nickkaltner on GitHub (May 27, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/10879 ### What is the issue? when i use codestral, the chat endpoint works fine with model and messages parameters. When i add the tools parameter, it 400s. I'm using the elixir ollama library, and it gives %Ollama.HTTPError{status: 400, message: "Bad Request"} here is the example script ```elixir model = "codestral:22b-v0.1-q2_K" stock_price_tool = %{ type: "function", function: %{ name: "get_stock_price", description: "Fetches the live stock price for the given ticker.", parameters: %{ type: "object", properties: %{ ticker: %{ type: "string", description: "The ticker symbol of a specific stock." } }, required: ["ticker"] } } } messages = [ %{role: "system", content: "You are a helpful assistant."}, %{role: "user", content: "what is apple's latest stock price?"} ] {:ok, resp} = Ollama.chat(client, model: model, messages: messages, tools: [stock_price_tool], keep_alive: 10000 ) IO.puts inspect(resp) IO.puts resp["message"]["content"] resp ``` If I comment out the tools: [stock_price_tool] it works without the tool use ### Relevant log output ```shell [GIN] 2025/05/27 - 21:19:14 | 400 | 20.177917ms | 127.0.0.1 | POST "/api/chat" [GIN] 2025/05/27 - 21:19:29 | 400 | 18.747583ms | 127.0.0.1 | POST "/api/chat" [GIN] 2025/05/27 - 21:26:01 | 400 | 18.512958ms | 127.0.0.1 | POST "/api/chat" ``` ### OS macOS ### GPU Apple ### CPU Apple ### Ollama version 0.7.1
GiteaMirror added the bug label 2026-05-04 17:28:37 -05:00
GiteaMirror commented 2026-05-04 17:28:40 -05:00
Author
Owner
Copy Link

@arturo-air commented on GitHub (May 27, 2025):

If you want to make tools calls, you should use models marked as tools. You probably want to try devstral, it is also from Mistral and it allows tools.

<!-- gh-comment-id:2912270527 --> @arturo-air commented on GitHub (May 27, 2025): If you want to make `tools` calls, you should use models [marked as tools](https://ollama.com/search?c=tools). You probably want to try [devstral](https://ollama.com/library/devstral), it is also from Mistral and it allows `tools`.
GiteaMirror commented 2026-05-04 17:28:41 -05:00
Author
Owner
Copy Link

@nickkaltner commented on GitHub (May 27, 2025):

I guess that's my confusion, why it's not marked as a tool. From the codestral page;

"Code generation is one of the most popular LLM use-cases, so we are really excited about the Codestral release. From our initial testing, it's a great option for code generation workflows because it's fast, has favorable context window, and the instruct version supports tool use. We tested with LangGraph for self-corrective code generation using the instruct Codestral tool use for output, and it worked really well out-of-the-box (see our video detailing this)."

-- Harrison Chase, CEO and co-founder of LangChain

https://mistral.ai/news/codestral

<!-- gh-comment-id:2914286721 --> @nickkaltner commented on GitHub (May 27, 2025): I guess that's my confusion, why it's not marked as a tool. From the codestral page; "Code generation is one of the most popular LLM use-cases, so we are really excited about the Codestral release. From our initial testing, it's a great option for code generation workflows because it's fast, has favorable context window, and the instruct version supports tool use. We tested with LangGraph for self-corrective code generation using the instruct Codestral tool use for output, and it worked really well out-of-the-box (see our [video detailing this](https://youtu.be/zXFxmI9f06M))." -- Harrison Chase, CEO and co-founder of LangChain https://mistral.ai/news/codestral
GiteaMirror commented 2026-05-04 17:28:42 -05:00
Author
Owner
Copy Link

@nickkaltner commented on GitHub (May 27, 2025):

(thanks for the devstral tip)

<!-- gh-comment-id:2914289013 --> @nickkaltner commented on GitHub (May 27, 2025): (thanks for the devstral tip)
GiteaMirror commented 2026-05-04 17:28:42 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jun 10, 2025):

The chat template for the original model doesn't include tool calling, which is likely why the ollama model doesn't have it. However, there are special tokens that are used for tool parsing, so somebody industrious enough could modify the template to add tool calling.

<!-- gh-comment-id:2958417542 --> @rick-github commented on GitHub (Jun 10, 2025): The [chat template](https://huggingface.co/mistralai/Codestral-22B-v0.1/blob/main/tokenizer_config.json#L6176) for the original model doesn't include tool calling, which is likely why the ollama model doesn't have it. However, there are [special tokens](https://huggingface.co/mistralai/Codestral-22B-v0.1/blob/main/tokenizer_config.json#L47) that are used for tool parsing, so somebody industrious enough could modify the template to add tool calling.
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#69208
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?