feat(bench): model picker for multi-model routers - #56
Open
kesslerio wants to merge 1 commit into
Open
Conversation
Adds a model picker dropdown to the benchmark dialog for multi-model routers like LiteLLM. When /v1/models returns multiple models, the benchmark uses the first model by default and the user can pick from the full list. - LlmProbe: detect multi-model routers, expose models array and benchmarkModel field - BenchmarkDialog: model picker dropdown with .bench-select styling - LlmPanel: pass benchmarkModel and models to BenchmarkDialog, use benchmarkModel in showcase URL params - types.ts: add benchmarkModel and models fields to LlmMetrics Re-submission of closed PR MiaAI-Lab#51, rebased on current upstream/main. Generated with [Devin](https://devin.ai) Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
kesslerio
force-pushed
the
feat/bench-model-picker
branch
from
August 19, 2026 22:29
ea5caa4 to
7113e98
Compare
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
TL;DR
When an LLM backend exposes multiple models via
/v1/models(e.g. LiteLLM routing to several backends), the benchmark dialog now shows a model picker dropdown instead of silently using the first model.What Problem This Solves
Multi-model routers like LiteLLM return a list of models from
/v1/models, but sparkDash's benchmark dialog only used the first one. If you wanted to benchmark a specific model behind the router, you had no way to pick it — the benchmark ran against whatever happened to be first in the list.Why This Change Was Made
I run LiteLLM in front of several models and needed to benchmark individual models behind the router. This PR adds model selection to the benchmark dialog so you can target any model the router exposes.
User Impact
/v1/modelsreturns 2+ modelsmodelIdRisk and Rollout
Low risk — the model picker only renders when
models.length > 1. Single-model backends hit the same code path as before. No migration needed; thebenchmarkModelandmodelsfields are optional onLlmMetricsand default tonull/undefined.Evidence
npm run typecheck— passes cleannpm run build— passes (57 modules, 424 KB JS bundle)npm test— 149 server tests pass, 0 failgit diff upstream/main..feat/bench-model-picker -- src/components/HarnessWizard.tsxis emptygrep -r "HarnessWizard\|SessionSource\|occupancyPoller" src/returns nothing on the bench branch