Repository navigation
Show output TPS in chat usage - #658
Merged
Merged
Conversation
Member
Author
|
@ericdallo I think this could be useful addition to the status line (next to the cost). |
Measure streaming responses from the first generation event to the usage update and expose the result as messageOutputTps. The window ends before tool execution so approval and tool time cannot lower the value. Non-streaming requests and providers without usage keep the field absent.
zikajk
force-pushed
the
add-output-tps-usage
branch
from
October 20, 2026 16:27
3b176b4 to
658f75f
Compare
Contributor
|
Gets my vote. I would enjoy this feature. |
Member
|
@zikajk yeah, just wanna avoid too many info in usage metrics, need to think the best way to show there but sounds good |
Member
Author
|
@ericdallo That reminds me that some providers (openAI in my case, > 20$ subs don't have 5h limit) shows |
Member
Author
|
@ericdallo Another option would be to show this next to the messages like we already do with time? |
Member
Just fixed this on master @zikajk will take a look at the PR + UX now |
ericdallo
approved these changes
Oct 25, 2026
ericdallo
added a commit
that referenced
this pull request
Oct 25, 2026
ChatGPT plans without a 5h limit report their weekly window in the primary slot and leave the secondary one empty: 0% used, without a duration and with a reset already past. It was still kept as a window, labeled with its slot name and shown as already reset, so clients rendered it with no value, e.g. `secondary — · 7d 37%` in eca-emacs. Skip slots that are unused and have no duration, from both the response headers and the wham/usage endpoint, as the Codex CLI does for the headers. Unused windows with a duration, like a fresh 5h one, are still shown. Reported in #658 (comment) 🤖 Generated with [ECA](https://eca.dev) (nubank-anthropic/opus-5.5 - max) Co-Authored-By: eca-agent <git@eca.dev>
Member
|
Implemented as this, I believe it's good enough for now, let's see 
|
ericdallo
added a commit
to editor-code-assistant/eca-emacs
that referenced
this pull request
Oct 25, 2026
The server now sends messageOutputTps in the usage content (editor-code-assistant/eca#658), the output tokens per second of the last model response, measured from its first generated token to its end, excluding tool runs and approval waits. Show it with a new :tps mode-line module, added to the default eca-chat-mode-line-format right after the elapsed time it joins, as in "⏱ 42s · 87 tok/s". Being its own module, it can be moved or removed; its leading separator is part of the new eca-chat-tps-format, to adapt it to the new place. It hides while the speed is unknown, as for non-streamed requests or after a compaction, when the server omits the field, and ignores the speed of subagents. 🤖 Generated with [ECA](https://eca.dev) (nubank-anthropic/opus-5.5 - max) Co-Authored-By: eca-agent <git@eca.dev>
|
Does the server send this new metric already for ollama, too? |
Member
|
@cpp977 we implemented in a generic way that get the output-tokens and calculate it so it should work, but LMK if any issues |
ericdallo
added a commit
to editor-code-assistant/eca-webview
that referenced
this pull request
Oct 26, 2026
The server now sends messageOutputTps in the usage content (editor-code-assistant/eca#658), the output tokens per second of the last model response, measured from its first generated token to its end, excluding tool runs and approval waits. Show it as its own "87 tok/s" pill in the chat subheader, as eca-emacs does with its own mode-line module, with a tooltip explaining what it measures. It hides while the speed is unknown, as for non-streamed requests, when the server omits the field. It comes after the usage so that in a narrow side panel, where it may not fit, it wraps to its own line instead of moving the context bar. 🤖 Generated with [ECA](https://eca.dev) (nubank-anthropic/opus-5.5 - max) Co-Authored-By: eca-agent <git@eca.dev>
ericdallo
added a commit
to editor-code-assistant/eca-vscode
that referenced
this pull request
Oct 26, 2026
…ader Shows the output speed of the last model response, e.g. "87 tok/s", that the server sends as messageOutputTps (editor-code-assistant/eca#658). 🤖 Generated with [ECA](https://eca.dev) (nubank-anthropic/opus-5.5 - max) Co-Authored-By: eca-agent <git@eca.dev>
ericdallo
added a commit
to editor-code-assistant/eca-intellij
that referenced
this pull request
Oct 26, 2026
…ader Shows the output speed of the last model response, e.g. "87 tok/s", that the server sends as messageOutputTps (editor-code-assistant/eca#658). 🤖 Generated with [ECA](https://eca.dev) (nubank-anthropic/opus-5.5 - max) Co-Authored-By: eca-agent <git@eca.dev>
ericdallo
added a commit
to editor-code-assistant/eca-desktop
that referenced
this pull request
Oct 26, 2026
…ader Shows the output speed of the last model response, e.g. "87 tok/s", that the server sends as messageOutputTps (editor-code-assistant/eca#658). Also brings the chat messages copy fix made for VS Code 1.143, a click-focusable messages scroller, harmless here. 🤖 Generated with [ECA](https://eca.dev) (nubank-anthropic/opus-5.5 - max) Co-Authored-By: eca-agent <git@eca.dev>
ericdallo
added a commit
to editor-code-assistant/eca-web
that referenced
this pull request
Oct 26, 2026
…ader Shows the output speed of the last model response, e.g. "87 tok/s", that the server sends as messageOutputTps (editor-code-assistant/eca#658). Also brings the chat messages copy fix made for VS Code 1.143, a click-focusable messages scroller, harmless here. 🤖 Generated with [ECA](https://eca.dev) (nubank-anthropic/opus-5.5 - max) Co-Authored-By: eca-agent <git@eca.dev>
|
It works fine, I was just running stale lisp compiled files in eca-emacs. Sorry for the noise. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Measure streaming responses from the first generation event to the usage update and expose the result as messageOutputTps. Non-streaming requests and providers without usage keep the field absent.