feat(inference): default to DeepSeek-V4-Pro model#435
Merged
Conversation
Change the default selected model on page load from DeepSeek-R1-0528 to DeepSeek-V4-Pro. Updates the GlobalFilterContext initial state and the PARAM_DEFAULTS used to omit redundant g_model from share URLs. - GlobalFilterContext: default selectedModel → Model.DeepSeek_V4_Pro - url-state PARAM_DEFAULTS.g_model → 'DeepSeek-V4-Pro' (so the new default is omitted from share links) - url-state tests: assert the new default and swap the omit/preserve fixtures accordingly - model-architecture e2e: pin to ?g_model=DeepSeek-R1-0528 explicitly, since R1 (not the new default DSv4-Pro, which has no architecture diagram entry) is what those tests exercise Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Changes the default selected model on page load from DeepSeek-R1-0528 to DeepSeek-V4-Pro (1.6T / 49B active).
Changes
GlobalFilterContext.tsx— initialselectedModelnow defaults toModel.DeepSeek_V4_Pro(still overridable viainitialModelprop /?g_model=URL param).url-state.ts—PARAM_DEFAULTS.g_model→'DeepSeek-V4-Pro', so the new default is omitted from share URLs (and the old R1 value is now serialized when explicitly selected).url-state.test.ts— assert the new default; swap the omit/preserve fixtures so they still test the intended behavior against the new default.model-architecture.cy.ts— pin the architecture-diagram suite to?g_model=DeepSeek-R1-0528explicitly. Those tests exercise R1's MoE + dense block diagram; DSv4-Pro has no entry inMODEL_ARCHITECTURES, so the diagram (correctly) renders nothing for it. Decoupling the suite from the app default keeps it valid.Notes
LEGACY_BARE_DEFAULT_MODEL_SLUG(compare-slug.ts) and the AI-chart invalid-input fallback still reference R1, so existing shared/comparelinks keep resolving.default-category model with data ingested, so it's a valid landing model.Verification
pnpm typecheck✅pnpm fmt✅vitest run src/lib/url-state.test.ts— 35 passed ✅🤖 Generated with Claude Code
Note
Low Risk
Localized default-model and URL-serialization change with no auth or data-path edits; legacy compare fallbacks for R1 are intentionally unchanged.
Overview
Inference now lands on DeepSeek V4 Pro instead of DeepSeek R1 when no
?g_model=orinitialModelis set. The same default is wired inGlobalFilterContext(Model.DeepSeek_V4_Pro) andPARAM_DEFAULTS.g_modelinurl-state.ts, so share URLs omitg_modelfor V4 Pro and include it when users explicitly pick R1.Tests follow the new default: Vitest asserts
DeepSeek-V4-Proand updates omit/preserve share-URL fixtures. The model architecture Cypress suite visits/inference?g_model=DeepSeek-R1-0528so MoE/diagram coverage stays on R1, which still has architecture data—V4 Pro does not.Reviewed by Cursor Bugbot for commit 61f766f. Bugbot is set up for automated code reviews on this repo. Configure here.