---
type: "post"
id: "routers-are-a-new-ai-primitive-13-minute-6b86f3"
slug: "routers-are-a-new-ai-primitive-13-minute-6b86f3"
title: "Routers Are a New AI Primitive (13 minute read)"
description: "Thesis: Model routers should be judged by verified work across long-running agent sessions, not only by their accuracy on isolated prompts."
retrieval_nugget: "Thesis: Model routers should be judged by verified work across long-running agent sessions, not only by their accuracy on isolated prompts."
published_at: "2026-10-15"
updated_at: "2026-10-15"
record_date: "2026-10-15"
date_kind: "published_at"
topics: ["prompts"]
source_urls: ["https://www.vals.ai/blogs/routers-new-ai-primitive"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:routers-are-a-new-ai-primitive-13-minute-6b86f3"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:e552aad01791fa2845a3bb9a738a7b9a3afd1cf99026dbba8ad3b116053a4503","reviewed_at":"2026-10-15T16:09:22.316Z","source_evidence_count":1,"verified_claim_count":2,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":2,"implication_count":1,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"Model routers should be judged by verified work across long-running agent sessions, not only by their accuracy on isolated prompts.","observed_facts":[{"text":"Vals reports that five different models occupy the top positions across its eight benchmarks, and that no model ranks in the top three on all of them.","source_urls":["https://www.vals.ai/blogs/routers-new-ai-primitive"]},{"text":"It calculates that an oracle router picking the best model for every task in hindsight would beat the best single model on the Vals Index by 14 points at about 40% lower cost.","source_urls":["https://www.vals.ai/blogs/routers-new-ai-primitive"]}],"mechanism":"A routing choice changes more than the price of one request. In a multi-step agent workflow, a weaker answer can trigger recovery work or alter later choices; switching models can also sacrifice prompt-cache savings.","why_now":"Vals says ten developers released 16 language models tracked by its index in the past month alone, increasing the selection burden for teams.","implications":["Teams evaluating routers should ask whether evidence captures task success, total spend, and session-level effects together, instead of relying on a single-call benchmark or provider-specific result."],"evidence_boundary":"The article describes Vals’ argument and planned comparison, alongside its own index and oracle calculation. It does not provide results from the proposed independent router comparison, and the oracle assumes hindsight-perfect choices that cannot be reproduced operationally.","watch_conditions":["Watch for Vals to publish results from the shared-harness comparison, including task domains, router configurations, quality measures, and total costs."],"related_records":[],"new_branch_reason":"No prior New Runtime analytical record is supplied for this distinct evaluation-methodology story."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/routers-are-a-new-ai-primitive-13-minute-6b86f3/","markdown":"https://newruntime.com/posts/routers-are-a-new-ai-primitive-13-minute-6b86f3.md","json":"https://newruntime.com/posts/routers-are-a-new-ai-primitive-13-minute-6b86f3.json"}
related_patterns: ["agent-economics-moves-to-completed-work"]
---

# Routers Are a New AI Primitive (13 minute read)

## Retrieval answer

Thesis: Model routers should be judged by verified work across long-running agent sessions, not only by their accuracy on isolated prompts.

Thesis: Model routers should be judged by verified work across long-running agent sessions, not only by their accuracy on isolated prompts. Vals AI’s post makes this case while announcing that it is building an independent comparison of router products.

Observed facts: Vals reports that five different models occupy the top positions across its eight benchmarks, and that no model ranks in the top three on all of them. It calculates that an oracle router picking the best model for every task in hindsight would beat the best single model on the Vals Index by 14 points at about 40% lower cost. The post explicitly treats this oracle as an upper-bound reference, not a real system result.

Mechanism: A routing choice changes more than the price of one request. In a multi-step agent workflow, a weaker answer can trigger recovery work or alter later choices; switching models can also sacrifice prompt-cache savings. Vals argues that an evaluation must therefore run live, long-horizon tasks and account for the cumulative consequences of decisions rather than score one choice against one public question.

Why now: Vals says ten developers released 16 language models tracked by its index in the past month alone, increasing the selection burden for teams. The source also notes that router providers have begun publishing evaluations, while claiming an independent comparison under common conditions is the next step.

First implication: Teams evaluating routers should ask whether evidence captures task success, total spend, and session-level effects together, instead of relying on a single-call benchmark or provider-specific result.

Evidence boundary: The article describes Vals’ argument and planned comparison, alongside its own index and oracle calculation. It does not provide results from the proposed independent router comparison, and the oracle assumes hindsight-perfect choices that cannot be reproduced operationally.

First watch condition: Watch for Vals to publish results from the shared-harness comparison, including task domains, router configurations, quality measures, and total costs.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Prompts](https://newruntime.com/topics/prompts.md): Explore the prompts topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)