Implement functionality similar to the OpenRouter Pareto Routing
Details:
https://openrouter.ai/openrouter/pareto-code
https://openrouter.ai/docs/guides/routing/routers/pareto-router
But I think we could do better and instead of the use of OpenRouter data we could use usage data published by OpenCode (Zen/Go) backend usage statistics, as OpenCode seems now more at the edge in terms of competetive LLM request handling than OpenRouter itself.
Actually, the use of Pareto calculated by Artificial Intelligence may be the best way to go:
https://artificialanalysis.ai/?models=claude-opus-5%2Cgpt-5-6-sol%2Cgpt-5-6-sol-xhigh%2Cgpt-5-6-sol-high%2Cgpt-5-6-sol-medium%2Cgpt-5-6-sol-low%2Cgpt-5-6-luna%2Cgpt-5-6-luna-medium%2Cgpt-5-6-luna-xhigh%2Cgpt-5-6-luna-high%2Cdeepseek-v4-flash%2Cgrok-4-5%2Cqwen3-8-max%2Ckimi-k3-low%2Ckimi-k3%2Ckimi-k2-6%2Ckimi-k2-7-code%2Cgemini-3-6-flash%2Cgemini-3-5-flash%2Cminimax-m3%2Cglm-5-2%2Cinkling
see section "Intelligence Index vs. Cost per Intelligence Index Task"
Implement functionality similar to the OpenRouter Pareto Routing
Details:
https://openrouter.ai/openrouter/pareto-code
https://openrouter.ai/docs/guides/routing/routers/pareto-router
But I think we could do better and instead of the use of OpenRouter data we could use usage data published by OpenCode (Zen/Go) backend usage statistics, as OpenCode seems now more at the edge in terms of competetive LLM request handling than OpenRouter itself.
Actually, the use of Pareto calculated by Artificial Intelligence may be the best way to go:
https://artificialanalysis.ai/?models=claude-opus-5%2Cgpt-5-6-sol%2Cgpt-5-6-sol-xhigh%2Cgpt-5-6-sol-high%2Cgpt-5-6-sol-medium%2Cgpt-5-6-sol-low%2Cgpt-5-6-luna%2Cgpt-5-6-luna-medium%2Cgpt-5-6-luna-xhigh%2Cgpt-5-6-luna-high%2Cdeepseek-v4-flash%2Cgrok-4-5%2Cqwen3-8-max%2Ckimi-k3-low%2Ckimi-k3%2Ckimi-k2-6%2Ckimi-k2-7-code%2Cgemini-3-6-flash%2Cgemini-3-5-flash%2Cminimax-m3%2Cglm-5-2%2Cinkling
see section "Intelligence Index vs. Cost per Intelligence Index Task"