---
title: 'In the News: November 10, 2026 (Morning)'
description: "Meta prices Muse Ember at $0.10 and $0.50 per million tokens as a cheap subagent model; its benchmarks are Meta's own."
canonical_url: 'https://darkfactory.dev/news/2026-11-10-morning'
markdown_url: 'https://darkfactory.dev/news/2026-11-10-morning.md'
collection: news
date_published: '2026-11-10T07:05:53-05:00'
date_modified: '2026-11-10T07:05:53-05:00'
---

# In the News: November 10, 2026 (Morning)


Meta released a small, cheap model on November 9 and is selling it as the subagent in a harness. The benchmarks are Meta's, and the one test of mergeable code favors Anthropic's small model. A second item is an AWS-published case of a production step moving from Sonnet to a small decision model, with a rollout method worth copying.

## 1. Meta ships Muse Ember, a $0.10 and $0.50 per million token model aimed at subagent work

**[Introducing Our Smallest, Most Efficient Model: Muse Ember](https://research.meta.ai/blog/muse-ember)** · Meta · November 9, 2026

Meta lists muse-ember-1.0 at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 200K tokens, with a 1M-token context window and up to 512K output tokens. Meta says it matches Claude Haiku 5.5's best Artificial Analysis Intelligence Index score at about a third of the cost, roughly $0.14 versus $0.41 per task. On coding, Meta's table has Ember ahead on Terminal-Bench 4.0 (40.9 against 39.2 for Haiku) and on Coding Agent Index-AA (46 against 37). Each model ran in its own best harness on Terminal-Bench: mini-swe-agent for Ember, Claude Code for Haiku. On FrontierCode 2.0-AA, which judges mergeability, Haiku leads 34.9 to 27.2. Meta's quoted partner, Cline's founder, calls it purpose-built for subagent work.

**Why it matters:** If you send test runs and targeted fixes to a cheaper subagent model, this is a new price point to try. The figures are the vendor's, and the only mergeability benchmark in Meta's own table favors Haiku, so measure it on your tasks before changing a harness default.

## 2. AWS reports replacing Sonnet 4.6 with a small decision model for ticket labeling, after a dual-run rollout

**[Amazon saves millions by replacing Sonnet 4.6 with Strands Decider](https://strandsagents.com/blog/amazon-saves-millions-replacing-sonnet-with-strands-decider/)** · Carr Onstott, Mehul Jaiswal, Strands Agents Team, AWS · November 9, 2026

Two Amazon engineers swapped Claude Sonnet 4.6 for Strands Decider, a model that returns a probability for each option in a fixed list instead of generating text, on an internal ticketing system with 20 label categories. They report average time per call falling from 4,215 ms to 897 ms and tokens per ticket from 6,724 to 2,010. At published list prices they put cost at about $0.0222 per ticket for Sonnet against about $0.0005 for Decider. The post says Decider beat production Sonnet on a separate set of their tickets but gives no accuracy figures. The rollout used a three-way config switch (Sonnet only, both in parallel, Decider only); in the parallel mode Sonnet's labels were applied while the two were compared. The authors call going back a 45-second config change. They also report that reordering JSON keys moved output probabilities by up to 0.19, so they now send sorted keys.

**Why it matters:** The authors say this only suits a closed choice from a known catalog at high volume. The part that carries over to agent pipelines is the parallel mode: run the replacement on live traffic, apply the incumbent's output, and judge every disagreement before cutting over.