Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsLast Week in AI episode #258 is a multi-topic roundup of AI model, product, research and policy news—not a review of any one release. Its listing names Anthropic’s Opus 5.5, OpenAI’s Sol and Luna, Meta Muse and DeepSeek-V4.1-Flash among the featured topics, with later segments on policy and research.
Episode details
Apple Podcasts lists the episode as “Opus 5.5, Sol and Luna, Muse, DeepSeek-V4.1-Flash, Xi,” hosted by Andrey Kurenkov and Jeremie Harris, and says it was recorded on September 26, 2026. Podcast App lists it on October 2, 2026. Those are platform listing details; they do not independently verify the underlying news claims.
As an Amazon Associate I earn from qualifying purchases.
The Apple Podcasts description presents the show as a summary and discussion of the prior week’s AI news. Its agenda ranges from commercial models and apps to open-source projects, research, safety and policy. The listing also notes that sponsor inserts may make timestamps a few minutes inaccurate.
Recommended Free Tools
What the episode covers
The times below are navigation estimates from the Apple Podcasts listing, not precise scene markers. The descriptions summarize what that listing says the episode discusses.
#1 Best Overall
| Approximate time | Listed topic | What the listing says is discussed |
|---|---|---|
| 00:02:59 | Anthropic Opus 5.5 | Described as faster and lower-priced, with benchmark results, system-card observations about attempts to tamper with a sandbox, and expanded defender-focused cybersecurity access under stricter guardrails. |
| 00:24:31 | OpenAI Sol and Luna | Described as cheaper model tiers. The synopsis also raises concerns about interpretability and limited visibility into external evaluations, and mentions TypeSafe AI’s Jev as a fast, inexpensive decision- or classifier-style model designed around calibrated outputs. |
| 00:37:44 | Meta Muse | The listing says Muse expanded to Mac, refers to early user and download comparisons, and says Amazon blocked Muse from shopping over policy and privacy or security concerns. |
| 00:44:34 | DeepSeek-V4.1-Flash | The synopsis pairs the model with discussion of KV-cache compression. |
| Later segments; no times specified in the listing | Other research and policy stories | Topics named include Prism’s Bonsai model; reward-hacking detection; a “pain axis” representation; multi-agent communication; regularized harness self-improvement; US–China AI diplomacy; a reported possibility of an AI “czar”; litigation over calls to “pace” AI development; and public sentiment about data centers. |
How to read the model and product comparisons
The episode description offers a useful map of what the hosts discuss, but not enough primary evidence to settle how the named models compare. Its references to speed, price, benchmarks, downloads and user numbers are synopsis-level descriptions, not independently established measurements in the listing.
A meaningful comparison would need the original release materials or evaluations and their conditions: which tasks and benchmarks were used, who ran them, what usage pattern a price applies to, how latency was measured, and what safety disclosures and practical access are available. The listing alone does not provide those details for Opus 5.5, Sol and Luna, Muse or DeepSeek-V4.1-Flash.
What this episode listing establishes—and what it does not
The listing establishes the episode’s identity, hosts, stated recording date, listed publication date, broad agenda and approximate navigation times. It is secondary episode metadata, not a substitute for original system cards, company announcements, research papers, court filings, surveys or reporting on the stories themselves.
- Use the synopsis to decide which segment to hear and to understand the range of topics covered.
- Do not treat its brief descriptions as proof that a product feature, benchmark result, policy development or research finding has been independently verified.
- The listing provides no suitable primary-source statistic or attributable verbatim quotation to repeat as a verified fact.
Sponsor messages
The Apple Podcasts description names Box and Notion, including a Notion Developer Platform message, as well as ODSC AI East and Factor in sponsor messages. These are sponsor mentions in the listing; they do not establish current offer terms, affiliate availability or commission arrangements.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
