ModelSpec is an analysis of alternatives for AI models: it decides which models fit a job from your requirements, sourced benchmarks and real cost, and shows its work.
Retrieval embeddings on a budget
Find the cheapest hosted embedding model without ignoring retrieval quality.
Worked example as of 2026-09-23, from the public data image.
Cheapest hosted embedding; retrieval still ranked.
Requirements
model.class = vectoriser: Retrieval embeddings require the model class that emits vectors.
Preferences
-offering.cost_per_task: Prefer the cheaper task; embedding a corpus is priced per token.retrieval: Keep retrieval quality so a cheap but poor embedder does not win on cost alone.
Applied template spec
spec_version- 1
where- model.class = vectoriser
optimizeweights-offering.cost_per_task- 0.6
retrieval- 0.4
Worked answer
Decision status: no_feasible.
No recommendation: the public image has no answer with enough evidence.
Unknown requirements
Some candidates may qualify. Missing facts:
offering.cost_per_task
Missing evidence and possible changes
- no model that meets the requirements has a value for offering.cost_per_task, retrieval
Decision dec_4021c9df09b9981aeab9d5dd on public snapshot snap_05a963d2477ac2fb.
What this template teaches
Cost leads, but a quality Prefer keeps the cheapest poor embedder from winning outright.
The live board uses current data. Its answer can change as evidence changes.
Human lookup is free on Decide. CLI, API, and MCP access requires a key. Access and pricing.