Anypick, open library for filtering LLMs based on capabilites and price

anypick is an open library for selecting the best large language model (LLM) across providers using filters over capabilities, pricing, and benchmarks, then picking a winner with a strategy. It only selects a model; the user wires the chosen model into their chat client. Two implementations share the same design, fixtures, and behavior: anypick-python and anypick-ts. The library supports two model providers: openrouter (the default) and vercel. openrouter is the only provider with both a model catalog and benchmarks, so it supports every strategy. vercel is models-only, so it must be paired with NoopBenchmarkObtainer, and only price-only strategies (cheapest) are meaningful. Filters reduce (models, scores) to a subset before a strategy picks a winner. They can be built ergonomically with ModelFilters or as composable predicates with pred.*. ModelFilters fields include include_ids, exclude_ids, include_makers, exclude_makers, min_prompt_price, max_prompt_price, min_completion_price, max_completion_price, min_expected_cost, max_expected_cost, expected_cost_weights, max_cache_read_price, min_context_length, max_context_length, modalities_in, modalities_exactly, excludes_modalities, output_modalities_in, output_modalities_exactly, excludes_output_modalities, requires_tools, requires_reasoning, requires_structured_outputs, min_benchmarks, and max_benchmarks. Predicate form allows negation, alternation, or custom logic by composing predicates with & (and), | (or), and ~ (not).