Kimi K3.1 Pro Lands 4th on Artificial Analysis as Claude Duet Enters Preview

This Week in LLMs, Monday, October 26, 2026. Covers October 19–25.

TL;DR

• Moonshot AI released Kimi K3.1 Pro on October 23. Artificial Analysis scores it 55 on its Intelligence Index, 4th overall and the top open-weights model once weights ship (Artificial Analysis).
• Anthropic opened a research preview of Claude Duet, a proactive agent for US Pro and Max subscribers, and made Fast Mode for Opus 5.5 generally available (Releasebot).
• Italy's rule requiring a human to make final AI-assisted hiring and firing decisions took effect October 23 (ActuIA).

Top Story: Kimi K3.1 Pro Puts Open Weights Near the Top

Moonshot AI launched Kimi K3.1 Pro on October 23. It is a mixture-of-experts model with 2.8 trillion total parameters, 104 billion active per request, and a 1 million token context window. API pricing is $2 per million input tokens and $10 per million output tokens, according to Notebookcheck's report.

Artificial Analysis lists it at 55 on its Intelligence Index and says it launched 4th and will lead open-weights models. Moonshot says full weights arrive by November 4. As of Sunday they were not on Hugging Face, so "open" is still a promise.

Moonshot also says it removed cyber-related data from training for the open release. On its own test, a checkpoint without that removal solved 90.5% of tasks, against 81% for the final model. Those figures have not been independently verified.

Why it matters: For business users, a near-frontier model at this price gives you leverage in vendor negotiations and a possible self-hosting option. For builders, wait for the weights and independent tests before moving workloads, and check the license.

Leaderboard Watch

Source: Artificial Analysis changelog. Claude Opus 5.5 still holds the top score on the Artificial Analysis leaderboard, at 58. We could not verify any LMArena rank changes this week, so none are reported.

Model & Pricing Changes

• Anthropic: Fast Mode for Opus 5.5 is generally available at $8 input and $40 output per million tokens (Releasebot).
• Anthropic: Claude on Google Cloud, managed by Anthropic, reached general availability on October 22, with usage counting toward Google Cloud commitments (Releasebot).
• Moonshot AI: Kimi K3.1 Pro API is priced at $2/$10 per million tokens, about a third below Kimi K3 (Notebookcheck).
• MiniMax: MiniMax-M3.1-Flash was evaluated by Artificial Analysis on October 19 at a score of 43 (Artificial Analysis). A third-party summary lists $0.15/$0.60 per million tokens (Local AI Zone); check MiniMax's price page before budgeting.
• Ling: Ling 3.1 Flash 1019 scored 44 on the same index (Artificial Analysis).

Policy Watch

• Italy: Legislative Decree 179/2026 requires a person with real decision power to make final AI-assisted hiring, evaluation, discipline and dismissal calls. Decisions made only by automated processing are void. The law applies from October 23. The decree text was not yet online when the cited analysis was written, so confirm details with counsel (ActuIA).
• Anthropic, coming November 12: The updated Usage Policy adds rules on deceptive campaigns, narrows election rules, and tightens limits on weapons and surveillance. Review it if you build on Claude (Anthropic).

Tools Worth a Look

• Claude Duet: A waitlist research preview of a background agent that asks approval before sending messages, paying or deleting things. It is limited to US Pro and Max users (Releasebot).
• Agent Skills donation: Anthropic is donating its Agent Skills open standard to the Agentic AI Foundation, which may make skills portable across vendors (Anthropic).
• Unified Claude for Enterprise: Chat and Cowork now run as one experience, with spend limits and pooled group budgets for admins (Releasebot).

Events (Next 30 Days)

• Oct 27–29: ODSC AI West, Burlingame, CA (Unite.AI)
• Nov 4: Moonshot's stated deadline for Kimi K3.1 Pro weights (Notebookcheck)
• Nov 4–5: Future of AI Summit, London (Unite.AI)
• Nov 10–12: Agentic AI Week USA, Atlanta (Unite.AI)
• Nov 12: Anthropic Usage Policy takes effect (Anthropic)
• Nov 16–17: AI Coding Summit NYC (Unite.AI)
• Nov 24–27: AI Summit Europe, Vilnius (Unite.AI)

Keep Up With the Rankings

Track every model as the numbers change at LLM1 Rankings, and get this roundup in your inbox through the newsletter at llm1.com.