All 5 AI packs for $39 →
← All posts
News · Models · Coding

This week in AI tools: GPT-6.1 Terra, Kimi K3.1 Pro, Claude Duet and Runway App Studio

Published · Facts checked against official vendor pages on October 27, 2026 · Rankings are independent of sponsorships

Two new models landed this week at prices worth a look: OpenAI's GPT-6.1 Terra and Moonshot's Kimi K3.1 Pro. Around them came a proactive agent app from Anthropic, a new home for the Agent Skills standard, and a way to build your own creative apps inside Runway. Here is what changed between October 21 and 26, 2026, what it costs, and what to do about it. Every price below was read from the vendor's own page on October 27, 2026.

Tools in this post

SChatGPTAOpenAI CodexAGitHub CopilotAOpenRouterBKimi (Moonshot AI)AGeminiSClaudeSClaude CodeSCursorARunway

The short version

GPT-6.1 Terra (Oct 26) is OpenAI's new fast, lower-cost model: $1 input / $5 output per 1M tokens in the API, in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu, and already GA in GitHub Copilot.

Kimi K3.1 Pro (Oct 23) is Moonshot's new open-weight model with a 1M-token context, at $2 / $10 per 1M tokens. Weights are due by Nov 4, and K2 models close to new users on Oct 31.

Claude Duet (Oct 21) is a waitlisted research preview of a proactive agent app for US Pro and Max subscribers.

Agent Skills (Oct 22), the SKILL.md format, now lives at the Linux Foundation's Agentic AI Foundation next to MCP and AGENTS.md.

Runway App Studio (Oct 26) builds and hosts custom creative apps from a plain-language description, on every paid plan.

GPT-6.1 Terra: what it costs and where you can use it

OpenAI pitches Terra as its fast, efficient model for computer use, visual reasoning and clear writing, and says it is a big step up from GPT-5.6 Terra on coding: 50.3% on Terminal-Bench 4.0 at max effort, against GPT-5.6 Terra's published 21.5%. It also says GPT-6.1 Sol and GPT-6 Astra are still the better choice for the most complex coding tasks. Treat these as OpenAI's own numbers.

API: the model ID is gpt-6.1-terra, on the Responses and Chat Completions APIs (tool calling needs Responses). It has a 1.05M-token context window and 128K max output. Prompts over 272K input tokens are billed at 2x input and 1.5x output for the whole request, and Batch and Flex are 50% off.

ChatGPT and Codex: Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. OpenAI says the same plan allowance covers more tasks with Terra than with GPT-5.6 Terra, GPT-6.1 Sol or GPT-6 Astra, so it is the model to pick when you keep hitting Codex limits on routine work. Plan prices are on our ChatGPT and OpenAI Codex pages.

GitHub Copilot: GitHub made Terra generally available on October 26 for Copilot Pro, Pro+, Max, Business and Enterprise, rolling out gradually across VS Code, Visual Studio, Copilot CLI, the cloud agent, JetBrains, Xcode and Eclipse. It is billed at provider list pricing from your AI Credits: GitHub's models page lists $1 input, $0.05 cached, $1.25 cache write and $5 output per 1M tokens up to 272K input tokens, and $2 / $0.10 / $2.50 / $7.50 above that. On Business and Enterprise, new models are enabled by default unless an admin has turned that off. See GitHub Copilot for what each plan includes.

Elsewhere: OpenRouter lists openai/gpt-6.1-terra at the same $1 / $5. Cursor's models page still listed GPT-5.6 Terra, not 6.1, when we checked on October 27.

How the new prices compare

Standard API list prices per 1M tokens (input / output), from each vendor's own docs on October 27, 2026. Benchmarks aside, this is the quickest way to see where Terra and K3.1 Pro sit:

  • GPT-6.1 Terra: $1 / $5 (cached input $0.05)
  • GPT-6.1 Sol: $2 / $10
  • GPT-5.6 Terra: $2 / $12
  • GPT-6 Astra: $10 / $50
  • Kimi K3.1 Pro: $2 / $10 (cached input $0.20); K3.1 Pro Highspeed: $4 / $20
  • Gemini 4 Argon: $2 / $10 through Nov 30, 2026, then $4 / $20 (paid tier only)

Kimi K3.1 Pro arrives, and K2 starts retiring

Moonshot released Kimi K3.1 Pro and a 2-3x faster Highspeed variant on October 23, across Kimi.ai, the Kimi apps, Kimi Work, Kimi Code and the Kimi API. It calls K3.1 Pro its most capable 3T-class open-weight model, and says full weights will be released by November 4, 2026. Moonshot reports 65.7% on Terminal-Bench 4.0 from its own run with a Claude Code harness.

In the API, K3.1 Pro always thinks (reasoning effort low, high or max, default max) and is unlocked after a minimum $1 top-up. The bigger practical change is on the pricing page: from October 31, 2026, K2 series models are no longer available to newly registered users, and Moonshot plans to retire all K2 models by the end of December 2026. If you run kimi-k2.6 or kimi-k2.7-code ($0.95 / $4), note that K3.1 Pro costs about 2 to 2.5 times as much per token ($2 / $10), and plan your switch before year-end. Our Kimi page has the updated prices.

Claude Duet and Agent Skills: agents that work while you're away

Claude Duet is a new, separate app (iOS, Android and claude.ai) where Claude, built around Opus 5.5, comes to you first. It sends a morning brief across your connected apps, checks them every few hours, and runs longer tasks in a virtual machine with its own browser. It asks for approval before sending messages, paying, signing or deleting anything. It is a waitlisted research preview for US Pro and Max subscribers, included in those plans, and its background checks count toward your usage limits. Plans are on our Claude page.

The day after, Anthropic donated Agent Skills to the Agentic AI Foundation, which already hosts MCP and AGENTS.md. A skill is a folder with a SKILL.md file plus any scripts it needs, and Anthropic says more than 70 platforms read the format, including ChatGPT, Cursor, GitHub Copilot and VS Code. The practical takeaway: project instructions and skills you write once now travel between agents. Start with our free AGENTS.md starter and CLAUDE.md template, read which instruction file each coding agent reads, and see the Claude Code Agents pack for ready-made subagents.

Runway App Studio: build your own creative tools

Runway's App Studio lets you describe a creative tool in plain language, then writes, tests, deploys and hosts it, using Runway's image, video and audio models plus your Brand Kits, Workflows and assets. You can turn an existing Agent session or Workflow into an app in one click, or start from built-in apps such as Character Swap, AutoComp and Product Portfolio.

It is included in every paid plan with no add-on. Building, running and hosting use your existing credits, and people who open a public app link pay for their own generations with their own Runway account. Free users can run built-in and shared apps but not build their own. Current plan prices are on our Runway page, and you can compare it with other tools in our best AI video generators list.

Smaller updates worth knowing

A few more changes from official changelogs this week:

  • GitHub Copilot assisted permissions are GA (Oct 26) in Copilot CLI, the Copilot app and VS Code: a model classifies each command, file or URL request, clearly safe ones run automatically, and the rest still ask you. Enterprise admins can block Allow All while keeping Assisted.
  • OpenAI's Decisions API is GA (Oct 22) on GPT-6 Luna, for fast classify, route and score calls, now with up to 400 questions per request.
  • Cursor's Agent SDK went GA on Oct 20: define an agent as a folder of code, test it with evals, and deploy it to Cursor or your own infrastructure. See Cursor.

What this means for our rankings

No tier changes this week. New models change what a tool can do, but we move a tool's tier only after a human review, not on launch day. Every item above has a dated, sourced entry in the changelog. If you're building agents on any of these models, the Agent Builder pack is model-agnostic, and our free support agent prompt works with any of them.

FAQ

How much does GPT-6.1 Terra cost?
In the OpenAI API, GPT-6.1 Terra costs $1 per 1M input tokens, $0.05 per 1M cached input tokens and $5 per 1M output tokens. Prompts over 272K input tokens cost 2x input and 1.5x output, and Batch and Flex are 50% off.
Is GPT-6.1 Terra in GitHub Copilot?
Yes. GitHub made it generally available on October 26, 2026 for Copilot Pro, Pro+, Max, Business and Enterprise, billed at provider list pricing under usage-based billing, with a gradual rollout.
Is Kimi K2 being discontinued?
Moonshot says K2 series models will no longer be available to newly registered users from October 31, 2026, and it plans to retire all K2 models by the end of December 2026. Its migration note points to Kimi K3.1 Pro.
Who can use Claude Duet?
Claude Duet is a waitlisted research preview, starting with Pro and Max subscribers in the US. It is included in those plans and shares their usage limits.

Free and paid rules for your agents

Related changelog

Sources

Prices change often. Always confirm on the vendor's site before buying. See the full tier lists and the changelog.

+ Your logo here