Add local models, cost-aware routing, and compute/cost awareness#8

Merged

rililive merged 16 commits into

mainfrom

claude/local-model-cost-optimization-nzo2q6

Jun 9, 2026

Commits on Jun 9, 2026

Add local models, cost-aware routing, and compute/cost awareness
claude
committed
Abstract model-selection policy into a ModelDecisionEngine
claude
committed
fix: merge conflicts
zizzle6717
committed
fix(ollama): make stream_options.include_usage best-effort

zizzle6717
and
claude
committed
fix(ollama): add an idle timeout so a hung model can't freeze the REPL

zizzle6717
and
claude
committed
fix(ollama): honor configured maxTokens

zizzle6717
and
claude
committed
docs(ollama): note the tool/text ordering assumption in toOpenAiMessages

zizzle6717
and
claude
committed
fix(ollama): emit a final SSE frame that lacks a trailing newline

zizzle6717
and
claude
committed
fix(router): stop escalating routine uses of common coding verbs

zizzle6717
and
claude
committed
fix(resources): base the RAM warning on total capacity, not free memory

zizzle6717
and
claude
committed
fix(ui): show 'cost unknown' for unpriced cloud models, not 'no API cost'

zizzle6717
and
claude
committed
fix(ui): don't label up-front routing as 'escalated'

zizzle6717
and
claude
committed
fix(loop): keep escalation sticky across follow-up turns

zizzle6717
and
claude
committed
fix: merge conflicts
zizzle6717
committed
fix: resolve broken import and regressed UI labeling
zizzle6717
committed
release: v0.2.0
zizzle6717
committed