github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #14680] OpenAPI spec: ChatStreamEvent missing usage fields in final chunk #56014

New Issue
Open
opened 2026-04-29 10:08:38 -05:00 by GiteaMirror · 6 comments
GiteaMirror commented 2026-04-29 10:08:38 -05:00
Owner
Copy Link

Originally created by @davidmigloz on GitHub (Mar 6, 2026).
Original GitHub issue: https://github.com/ollama/ollama/issues/14680

Description

The OpenAPI spec at docs/openapi.yaml defines ChatStreamEvent with only 4 fields: model, created_at, message, done.

However, the Ollama API sends additional usage/timing fields in the final streaming chunk (when done: true), just like it does for GenerateStreamEvent and ChatResponse. These fields are missing from the ChatStreamEvent schema:

  • done_reason
  • total_duration
  • load_duration
  • prompt_eval_count
  • prompt_eval_duration
  • eval_count
  • eval_duration

Current state

Schema Has usage fields?
GenerateResponse ✅
GenerateStreamEvent ✅
ChatResponse ✅
ChatStreamEvent ❌ (missing)

Expected

ChatStreamEvent should include the same usage fields as GenerateStreamEvent, since the API returns them in the final chunk.

Impact

SDK generators that rely on the OpenAPI spec produce types that can't capture usage metrics from chat streaming responses, forcing users to bypass the typed API and parse raw NDJSON.

Originally created by @davidmigloz on GitHub (Mar 6, 2026). Original GitHub issue: https://github.com/ollama/ollama/issues/14680 ## Description The OpenAPI spec at `docs/openapi.yaml` defines `ChatStreamEvent` with only 4 fields: `model`, `created_at`, `message`, `done`. However, the Ollama API sends additional usage/timing fields in the **final streaming chunk** (when `done: true`), just like it does for `GenerateStreamEvent` and `ChatResponse`. These fields are missing from the `ChatStreamEvent` schema: - `done_reason` - `total_duration` - `load_duration` - `prompt_eval_count` - `prompt_eval_duration` - `eval_count` - `eval_duration` ## Current state | Schema | Has usage fields? | |---|---| | `GenerateResponse` | ✅ | | `GenerateStreamEvent` | ✅ | | `ChatResponse` | ✅ | | `ChatStreamEvent` | ❌ (missing) | ## Expected `ChatStreamEvent` should include the same usage fields as `GenerateStreamEvent`, since the API returns them in the final chunk. ## Impact SDK generators that rely on the OpenAPI spec produce types that can't capture usage metrics from chat streaming responses, forcing users to bypass the typed API and parse raw NDJSON.
GiteaMirror added the documentation label 2026-04-29 10:08:38 -05:00
GiteaMirror commented 2026-04-29 10:08:41 -05:00
Author
Owner
Copy Link

@davidmigloz commented on GitHub (Mar 6, 2026):

For reference, the Go code uses a single ChatResponse struct for both streaming and non-streaming responses (api/types.go), which includes:

  • Embedded Metrics struct (total_duration, load_duration, prompt_eval_count, prompt_eval_duration, eval_count, eval_duration)
  • done_reason
  • logprobs

So the OpenAPI spec's ChatStreamEvent should match ChatResponse (minus any non-streaming-only fields), just like GenerateStreamEvent matches GenerateResponse.

logprobs is also missing from ChatStreamEvent in the spec but present in GenerateStreamEvent.

<!-- gh-comment-id:4014789456 --> @davidmigloz commented on GitHub (Mar 6, 2026): For reference, the Go code uses a single `ChatResponse` struct for both streaming and non-streaming responses ([api/types.go](https://github.com/ollama/ollama/blob/main/api/types.go)), which includes: - Embedded `Metrics` struct (`total_duration`, `load_duration`, `prompt_eval_count`, `prompt_eval_duration`, `eval_count`, `eval_duration`) - `done_reason` - `logprobs` So the OpenAPI spec's `ChatStreamEvent` should match `ChatResponse` (minus any non-streaming-only fields), just like `GenerateStreamEvent` matches `GenerateResponse`. `logprobs` is also missing from `ChatStreamEvent` in the spec but present in `GenerateStreamEvent`.
GiteaMirror commented 2026-04-29 10:08:41 -05:00
Author
Owner
Copy Link

@kanusahai commented on GitHub (Mar 8, 2026):

Hello, can I take this up ?

<!-- gh-comment-id:4018367716 --> @kanusahai commented on GitHub (Mar 8, 2026): Hello, can I take this up ?
GiteaMirror commented 2026-04-29 10:08:42 -05:00
Author
Owner
Copy Link

@Rishi-11-2 commented on GitHub (Mar 8, 2026):

Hi I have opened a Pull Request for the same .
can you review it : https://github.com/ollama/ollama/pull/14717/commits

<!-- gh-comment-id:4018932420 --> @Rishi-11-2 commented on GitHub (Mar 8, 2026): Hi I have opened a Pull Request for the same . can you review it : https://github.com/ollama/ollama/pull/14717/commits
GiteaMirror commented 2026-04-29 10:08:42 -05:00
Author
Owner
Copy Link

@Rishi-11-2 commented on GitHub (Mar 9, 2026):

@davidmigloz please review my pr and do I need to anything else or improve it. Thank you

<!-- gh-comment-id:4023288826 --> @Rishi-11-2 commented on GitHub (Mar 9, 2026): @davidmigloz please review my pr and do I need to anything else or improve it. Thank you
GiteaMirror commented 2026-04-29 10:08:43 -05:00
Author
Owner
Copy Link

@kanusahai commented on GitHub (Mar 10, 2026):

It seems like @Rishi-11-2 got to it sooner. His change LGTM but I'll let @davidmigloz review.

<!-- gh-comment-id:4028485042 --> @kanusahai commented on GitHub (Mar 10, 2026): It seems like @Rishi-11-2 got to it sooner. His change LGTM but I'll let @davidmigloz review.
GiteaMirror commented 2026-04-29 10:08:43 -05:00
Author
Owner
Copy Link

@davidmigloz commented on GitHub (Mar 10, 2026):

Thanks for the PR! I've posted my feedback there

<!-- gh-comment-id:4029799194 --> @davidmigloz commented on GitHub (Mar 10, 2026): Thanks for the PR! I've posted my feedback there
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 documentation
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#56014
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?