Raise hosted context window 64K -> 128K#30
Merged
Merged
Conversation
The banner's '65.5K' came from a flat context_length: 65536 in the config (the model block + the opengradient provider), not from the model -- so the harness compacted at 64K even though every hosted model handles far more. DeepSeek V4 Pro and Hermes 4 405B/70B are all >=128K native, so bump both hosted entries to 131072. Conservative on purpose (won't over-send past the gateway); can go higher if the gateway serves the full window. Local (ollama) windows unchanged.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The status bar showed
61.2K/65.5Kbecause the config hard-setcontext_length: 65536(themodelblock + theopengradientprovider) -- a flat 64K default unrelated to the model. So the harness compacted at 64K even though the hosted models handle much more.DeepSeek V4 Pro and Hermes 4 405B/70B are all >=128K native, so this bumps both hosted entries to
131072(128K). Left the local ones (ollama_num_ctx,ollama-local) alone -- the abliterated local models have small windows.Conservative on purpose: set to the safe native floor so we never send past the gateway's cap. If the OG gateway serves DeepSeek's full window (up to 1M), this can go higher.