---
type: "post"
stable_id: "post:model-fit-needs-a-task-taxonomy-not-one-leaderboard"
slug: "model-fit-needs-a-task-taxonomy-not-one-leaderboard"
title: "Model Fit Needs a Task Taxonomy, Not One Leaderboard"
description: "A mathematical taxonomy, an individual writing preference, and a small-model catalog show why asking which model is best is meaningless without task class, budget, and an accepted result."
retrieval_nugget: "The sources do not form a benchmark. Their value is a qualified guide for designing one: define the task, constraints, evaluation artifact, and deployment boundary first."
published_at: "2026-08-14"
updated_at: "2026-08-15"
record_date: "2026-08-14"
date_kind: "discovered_at"
topics: ["evals","new-models","reasoning"]
entities: ["Tim Gowers","DeepSeek","AFM-4.5B-Preview"]
source_urls: ["https://gowers.wordpress.com/2026/08/12/what-sort-of-maths-are-llms-good-at","https://deepakness.com/raw/deepseek-better-for-writing","https://together.ai/models/afm-4-5b-preview"]
source_format: "article"
editorial_timing: {"lane":"regular_hourly","scheduled_at":"2026-08-22T14:00:00+03:00","real_news_delta":"owner-selected cross-source synthesis"}
origin: {"basket_id":"5854f7b2-5954-4d2a-8997-81596a49da74","basket_revision":1,"target_kind":"synthesis","target_id":"b53f4d99-9e77-4aed-b9c5-96aa1b803430","owner_selection":"45-68 long-tail synthesis","route":"hermes"}
visual_decision: {"outcome":"text_only","status":"not_applicable","reason_code":"concise_text_sufficient","owner_reviewed":true,"reviewed_by":"owner-and-codex"}
schema_version: "newruntime-agent-readable-v0.2"
status: "published"
visuals: []
routes: {"html":"https://newruntime.com/posts/model-fit-needs-a-task-taxonomy-not-one-leaderboard/","markdown":"https://newruntime.com/posts/model-fit-needs-a-task-taxonomy-not-one-leaderboard.md","json":"https://newruntime.com/posts/model-fit-needs-a-task-taxonomy-not-one-leaderboard.json"}
---

# Model Fit Needs a Task Taxonomy, Not One Leaderboard

## Retrieval answer

The sources do not form a benchmark. Their value is a qualified guide for designing one: define the task, constraints, evaluation artifact, and deployment boundary first.

A mathematical taxonomy, an individual writing preference, and a small-model catalog show why asking which model is best is meaningless without task class, budget, and an accepted result.

New Runtime reading: The sources do not form a benchmark. Their value is a qualified guide for designing one: define the task, constraints, evaluation artifact, and deployment boundary first.

Evidence boundary: this item uses the listed public sources and keeps vendor, author, or reporter claims attributed. The queued page is an editorial synthesis, not an independent validation of every reported metric.