Repository navigation
feat(guardrails): add DeepSeek V4.1 Flash, retire V4 ids - #337
Merged
Merged
Conversation
DeepSeek's 2026-09-10 announcement (deepseek.com/en/news/ deepseek-v4-1-flash): deepseek-flash (V4.1-Flash, 552B MoE, 1M ctx / 384k out, native image input) is live on the API; deepseek-v4-flash and deepseek-v4-flash-vision-exp are retired (temp-routing to V4.1-Flash) and deepseek-v4-pro is phased out — from 2026-09-14 04:00 UTC all deepseek-v4-pro requests route to V4.1-Flash at V4.1-Flash rates, i.e. Pro-labeled billing for Flash behavior. - whitelist (both copies) + paid set + catalog snapshot now admit deepseek-chat, deepseek-flash, deepseek-reasoner; the three retired V4 ids are removed from all three copies together (the sync check requires snapshot ⊆ whitelist, so they move as a set) - snapshot refreshed from the live catalog (models.opencode.ai, 2026-09-12): also adds openai gpt-6-astra and 22 new openrouter models per the add-only policy; one openrouter granite id that left the catalog stays whitelisted (add-only protects catalog-unlisted routes) - project .opencode/opencode.jsonc model rows: v4-flash/v4-pro → deepseek-flash - cor-local's deepseek-v4-flash-0731 is the Mac Studio llama-server GGUF name, not an API id — untouched Agent-Lane: opencode
|
This PR doesn't fully meet our contributing guidelines and PR template. What needs to be fixed:
Please edit this PR description to address the above within 2 hours, or it will be automatically closed. If you believe this was flagged incorrectly, please let a maintainer know. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Stacks on #336 (upstream merge, merged). Branch is based on the upstream-merge content; diff vs current dev is exactly the DeepSeek admission change.
Summary
DeepSeek's 2026-09-10 announcement:
deepseek-flash(V4.1-Flash) live on the API with native multimodal input;deepseek-v4-flash/deepseek-v4-flash-vision-expretired;deepseek-v4-prophased out (from 2026-09-14 04:00 UTC it routes to V4.1-Flash at V4.1-Flash rates — Pro-labeled billing for Flash behavior).paidset, catalog snapshot): nowdeepseek-chat,deepseek-flash,deepseek-reasoner. The three retired V4 ids removed from all three copies together (sync check requires snapshot ⊆ whitelist).gpt-6-astra+ 22 openrouter models added per add-only policy..opencode/opencode.jsoncproject model rows updated.cor-local'sdeepseek-v4-flash-0731is the Mac Studio llama-server GGUF name — untouched.H5-E2E: python3 scripts/sync-model-whitelist.py && (cd packages/guardrails && bun test profile/plugins)
H5-E2E-OUT: ok whitelist covers the catalog snapshot in all three copies; 79 pass / 0 fail, 142 expect() calls across 7 files [1.72s]
H5-SUBTRACTION: N/A
H5-PAIR: N/A