Kimi K3.1 Pro: what it is and what Moonshot AI says it does

Kimi K3.1 Pro is an open-weight AI model from Moonshot AI. It is built for coding, 3D creation, financial research and agentic work, and has a 1-million-token context window.

Key facts

• Maker: Moonshot AI
• Announced/released: the Unite.AI report, published on 23 October 2026, says Moonshot AI introduced the model on 24 October 2026, while OpenRouter lists a release date of 23 October 2026. The two sources therefore differ by a day.
• Price: OpenRouter lists $2 / $10 per 1M tokens. No GBP price is given.
• Status: available today on Kimi.ai, Kimi Work, Kimi Code and the Kimi API. Moonshot AI says the full model weights will be released by 4 November.

What is Kimi K3.1 Pro?

Kimi K3.1 Pro is the first model in Moonshot AI's new K3.1 series. The company describes it as its most capable 3T-class open-weight model to date, with a 1-million-token context window. "Open-weight" means the trained model weights are published so that others can download and run the model. In this case the maker says the full weights will follow by 4 November rather than being out already.

It is a post-trained version of Kimi K3, built with what Moonshot AI describes as substantially larger post-training compute. To enable large-scale joint reinforcement learning across hundreds of downstream domains, the company says it optimised every layer of the stack: the reinforcement learning framework, long-context inference, training efficiency, sandbox infrastructure, rollout scheduling and a shared platform for environment quality checks. Reinforcement learning is a method in which a model improves by being scored on its results.

OpenRouter describes it as Moonshot AI's flagship open-weight multimodal reasoning model and a post-trained successor to Kimi K3. It says the model is built for long-horizon coding, 3D creation, financial research and end-to-end agentic knowledge work, with native image and video understanding.

A faster companion, Kimi K3.1 Pro Highspeed, is also available. Moonshot AI says it delivers two to three times the speed of K3.1 Pro, and the Unite.AI report says it does so with identical model capabilities. The company positions it for more responsive interactions.

What can Kimi K3.1 Pro do?

Moonshot AI says the model delivers substantial improvements across coding, 3D creation, financial research and other agentic capabilities. Those are the same areas OpenRouter lists as the model's focus: long-horizon coding, 3D creation, financial research and end-to-end agentic knowledge work.

The company's emphasis on long-horizon work is reflected in how it describes the model's training. The large-scale joint reinforcement learning across hundreds of downstream domains is described by the company alongside an internal benchmark, Kimi Code Bench 3.0 Long-horizon, covering multi-step execution, tool use and end-to-end problem solving.

The 1-million-token context window matters for this kind of work. A context window is the amount of text a model can consider at once, and a larger one lets the model keep more of a codebase, a set of documents or a long task history in view. OpenRouter also lists the context as 1.0M.

The "Highspeed" variant is aimed at a different need. For interactive use, where a user is waiting on each response, speed matters as much as raw capability, and Moonshot AI's pitch is that Highspeed offers the same capabilities at two to three times the speed.

How does Kimi K3.1 Pro rank against other models?

The sources give a handful of benchmark scores, reported by Moonshot AI and repeated by Unite.AI.

On Terminal-Bench 4.0, K3.1 Pro scored 65.7%, ahead of GPT-6 Astra at 58.2% and narrowly behind Claude Opus 5.5 at 66.4%. That puts it narrowly behind the leader in the comparison given.

Kimi Code Bench 3.0 Long-horizon is described as an internal benchmark covering multi-step execution, tool use and end-to-end problem solving. Because it is the company's own test, it is best read as an indication of what Moonshot AI is optimising for rather than as an independent ranking.

How do you use Kimi K3.1 Pro?

Kimi K3.1 Pro and Kimi K3.1 Pro Highspeed are available today in several places:

• Kimi.ai and the Kimi mobile apps for iOS, Android and HarmonyOS.
• Kimi Work, the desktop application, version 3.1.0 or later, for Windows and Apple silicon Macs.
• Kimi Code, where users select the model with the /model command.
• The Kimi API, where the models are listed as kimi-k3.1-pro and kimi-k3.1-pro-highspeed. Zero Data Retention is available for API customers.
• OpenRouter, a third-party routing service, lists the model as moonshotai/kimi-k3.1-pro.

Kimi Enterprise administrators can enable K3.1 Pro for their organisations, with access off by default at launch.

What does it cost?

OpenRouter lists the model at $2 / $10 per 1M tokens, with a context of 1.0M. OpenRouter also explains that the average price customers actually pay is shown next to the prices providers post, and that caching and discounts mean the price actually paid is often well below the listed one.

OpenRouter routes requests to providers according to the routing mode a user picks: Balanced (price and speed), Nitro (fastest) or Floor (cheapest). Throughput, latency, time to first token, uptime and availability are tracked on the model's page, so anyone choosing a provider can compare them there.

What does this mean for UK and US readers?

On pricing, the only published figures are in US dollars, via OpenRouter. No GBP price is given, so UK readers should check the providers directly rather than assume a conversion.

The Zero Data Retention option for API customers may matter to organisations with data-handling concerns, and Kimi Enterprise administrators can control whether K3.1 Pro is enabled for their organisation. The model is available through Kimi's apps, desktop client, coding tool and API, so access does not depend on a single route.

What is verified vs claimed?

Listed by a third party (OpenRouter), not tested independently:

• OpenRouter lists the model as Moonshot AI's open-weight multimodal reasoning model with a 1M-token context window.
• OpenRouter lists a price of $2 / $10 per 1M tokens and a release date of 23 October 2026.

OpenRouter's description largely matches Moonshot AI's own wording, so it is a listing rather than an independent evaluation.

Claimed by Moonshot AI, the maker:

• The Terminal-Bench 4.0 score of 65.7% and the comparisons with GPT-6 Astra at 58.2% and Claude Opus 5.5 at 66.4%.
• The results on its internal Kimi Code Bench 3.0 Long-horizon benchmark, which is the company's own test.
• That Highspeed delivers two to three times the speed of the standard model, with identical model capabilities according to the Unite.AI report.
• That the model brings substantial improvements across coding, 3D creation, financial research and other agentic capabilities.
• That the full weights will be released by 4 November. Until then, this is a stated intention.

Unite.AI's article largely repeats Moonshot AI's announcement, so it should be read as a report of the company's claims rather than independent confirmation.

Related on Future Technology - LLM VRAM calculator: work it out for your own hardware - Moonshot AI paused Kimi K3 signups because it ran out of capacity - Claude Sonnet 5 is now the default model for every Free and Pro user - Touchscreen MacBook Pro on 28 October? Why you should wait to buy - All explainers

Frequently asked questions ### What is Kimi K3.1 Pro?

It is Moonshot AI's flagship open-weight multimodal reasoning model, a post-trained successor to Kimi K3. It has a 1-million-token context window and is built for long-horizon coding, 3D creation, financial research and agentic work.

How much does Kimi K3.1 Pro cost?

OpenRouter lists $2 / $10 per 1M tokens. No GBP price is given.

When will the Kimi K3.1 Pro weights be released?

Moonshot AI says the full model weights will be released by 4