WildDragonLLC/claudecodeui
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat: add per-model context window overrides

Browse source 
Adds MODEL_CONTEXT_OVERRIDES map and getModelContextWindow() helper so
local/constrained models (claudecode, cc/qwen72b, ollama-local/*) declare
their VRAM-limited context budget. claude-sdk.js will stamp this into
the subprocess env so Claude Code self-limits instead of overflowing the model.
This commit is contained in:
Zac Gaetano 2026-05-31 17:18:49 -04:00
parent 25ca157f9d
commit 69e478224c
1 changed files with 33 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

33
shared/modelConstants.js
View file

@ -48,6 +48,8 @@ export const CLAUDE_MODELS = {
{ value: "ollama/glm-4.7-flash", label: "GLM-4.7 Flash (ollama)" }, { value: "ollama/glm-4.7-flash", label: "GLM-4.7 Flash (ollama)" },
{ value: "ollama/qwen3.5", label: "Qwen 3.5 (ollama)" }, { value: "ollama/qwen3.5", label: "Qwen 3.5 (ollama)" },
{ value: "ollama-local/r1-14b-32k", label: "R1 14B 32K (local)" }, { value: "ollama-local/r1-14b-32k", label: "R1 14B 32K (local)" },
{ value: "claudecode", label: "claudecode (local)" },
{ value: "cc/qwen72b", label: "qwen72b (local)" },
{ value: "gemini/gemini-3.1-pro-preview", label: "Gemini 3.1 Pro Preview" }, { value: "gemini/gemini-3.1-pro-preview", label: "Gemini 3.1 Pro Preview" },
{ value: "gemini/gemini-3.1-flash-lite-preview", label: "Gemini 3.1 Flash Lite" }, { value: "gemini/gemini-3.1-flash-lite-preview", label: "Gemini 3.1 Flash Lite" },
{ value: "gemini/gemini-3-flash-preview", label: "Gemini 3 Flash Preview" }, { value: "gemini/gemini-3-flash-preview", label: "Gemini 3 Flash Preview" },
@ -104,6 +106,37 @@ export const CLAUDE_MODELS = {
DEFAULT: "cc/claude-sonnet-4-6", DEFAULT: "cc/claude-sonnet-4-6",
}; };
/**
* Per-model context window overrides (in tokens).
*
* Models listed here will have CONTEXT_WINDOW stamped into the Claude Code
* subprocess env so the agent self-limits its prompt budget accordingly.
* Models NOT listed here fall back to the global CONTEXT_WINDOW env var (default 160000).
*
* Add an entry here whenever you add a local/constrained model whose VRAM or
* architecture limits the usable context below the global default.
*/
export const MODEL_CONTEXT_OVERRIDES = {
'claudecode': 60000,
'cc/qwen72b': 60000,
'ollama-local/r1-14b-32k': 32000,
};
/**
* Returns the effective context window size for a given model value.
* Checks MODEL_CONTEXT_OVERRIDES first, then falls back to the
* global CONTEXT_WINDOW env var, then 160000.
*
* @param {string} modelValue - The model value string (e.g. "claudecode", "cc/claude-sonnet-4-6")
* @returns {number} Context window in tokens
*/
export function getModelContextWindow(modelValue) {
if (modelValue && MODEL_CONTEXT_OVERRIDES[modelValue] !== undefined) {
return MODEL_CONTEXT_OVERRIDES[modelValue];
}
return parseInt(process.env.CONTEXT_WINDOW, 10) || 160000;
}
/** /**
* Cursor Models * Cursor Models
*/ */