fix(agent): add missing BytePlus ModelArk catalog entries - #50
Open
leksautomate wants to merge 1 commit into
Open
fix(agent): add missing BytePlus ModelArk catalog entries#50leksautomate wants to merge 1 commit into
leksautomate wants to merge 1 commit into
Conversation
deepseek-v3-2-251201 and seed-2-0-pro-260328 were absent from the
hand-vendored BYTEPLUS_MODELS catalog (BytePlus has no models.dev
provider), so resolveModelCapabilities() fell back to the
unknown-model defaults: an 8,192-token context window and a
2,048-token output cap. That fallback context window is far below
what OpenChatCut's own system prompt and tool schemas need, so
context-compaction.ts's own trigger correctly refused to send the
request at all ("The current request is too large for this model
context window"), on every turn, regardless of conversation size.
Verified the real limits directly against the account's ModelArk
endpoint (GET /api/v3/models, which returns per-model token_limits):
deepseek-v3-2-251201 supports a 131,072-token context window with a
32,768-token output cap, and seed-2-0-pro-260328 supports 262,144
context with a 131,072 output cap. Added both to the catalog and to
the generator script that rebuilds it, matching the existing
hand-vendored ModelArk entries.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
deepseek-v3-2-251201andseed-2-0-pro-260328(BytePlus ModelArk) were missing from the hand-vendoredBYTEPLUS_MODELScatalog inscripts/update-llm-model-capabilities.mjs/assets/model-capabilities/models-dev.json(BytePlus has nomodels.devprovider, so these entries are hand-maintained).resolveModelCapabilities()fell back to the unknown-model defaults (8,192 context / 2,048 output). That fallback context window is smaller than OpenChatCut's own system prompt + tool schemas, socontext-compaction.tscorrectly refused to send the request at all ("The current request is too large for this model context window...") on every single turn, regardless of conversation length.GET /api/v3/models, which returns per-modeltoken_limits):deepseek-v3-2-251201→ 131,072 context / 32,768 max output;seed-2-0-pro-260328→ 262,144 context / 131,072 max output. Added both to the catalog JSON and to the generator script so they survive the nextmodels.devresync.Test plan
npx tsc --noEmit— cleannpx oxlint scripts/update-llm-model-capabilities.mjs— cleannode -e "JSON.parse(...)"— catalog JSON still validresolveModelCapabilities({ backend: 'api', provider: 'byteplus', modelId })for both new model IDs — resolves to the real vendor values withsource: 'catalog',estimated: false; a genuinely unknown model id still correctly falls back to the conservative defaultstsx src/agent/ai-sdk.verify.ts,tsx src/agent/context-compaction.verify.ts,tsx src/agent/model-selection.verify.ts,tsx src/agent/local-models.verify.ts,tsx src/agent/vision.verify.ts,tsx server/keystore.verify.ts— all pass