Kimi K3.1 Pro: What's New, Finance Workflows, and Alternatives (2026)
Kimi K3.1 Pro is the newest flagship model from Moonshot AI. It launched on October 23, 2026, in the Kimi app, Kimi Work, Kimi Code, and the Kimi API. Moonshot calls it its "most capable 3T-class open-weight model to date," with a context window of one million tokens. The full weights are due by November 4.
This guide covers what changed, with a close look at the finance work Moonshot highlights. It also explains how to access the model and what the API costs. The last sections compare the main alternatives. Every detail comes from the vendors' own pages, read on October 26, 2026.
What Moonshot announced on October 23
Moonshot's launch post opens with one sentence that sets the scope. Kimi K3.1 Pro "delivers substantial improvements across coding, 3D creation, financial research, and other agentic capabilities."
The release has five main parts:
- A new model series. Kimi K3.1 Pro is "the first model in our new K3.1 series." Moonshot says it is "a post-trained version of Kimi K3, with substantially larger post-training compute."
- A faster variant. Kimi K3.1 Pro Highspeed "delivers 2-3× the speed of K3.1 Pro for more responsive interactions."
- Open weights, later. The post says "Full model weights will be released by Nov 4, 2026." Until then, you use the model as a hosted service, through Kimi's apps or an API.
- Wide availability on day one. The model "rolls out today across the Kimi product family." That covers Kimi.com, the Kimi mobile apps, Kimi Work, Kimi Code, and the Kimi API.
- A new 3D feature. Moonshot says "3D Make is now live on kimi.com and the Kimi app." It lets you preview and edit 3D models, then download them for 3D printing.
Moonshot's official account posted the launch on X on October 23, 2026, a Friday. Simon Willison's blog confirmed the same day that the model had "landed on the Kimi API today."
What was already there before
Kimi K3.1 Pro builds on Kimi K3, so it helps to know the base.
Moonshot's Kimi K3 post describes "a 2.8T-parameter model" with "native vision capabilities and a 1-million-token context window." It calls K3 "the world's first open 3T-class" model. Kimi K3.1 Pro keeps that base and adds more training on top.
The Kimi product family was also in place. Kimi Work is a desktop app for Windows and Apple silicon Macs. Kimi Code is the coding tool. The Kimi API serves developers. Kimi Business "provides enterprise-grade data privacy and member management."
What is new is the training focus. Moonshot says it ran "large-scale joint RL across hundreds of downstream domains." Finance, 3D, and long coding tasks got the most attention in the launch post.
Finance workflows: what Kimi K3.1 Pro is built to deliver
Finance gets its own section in the launch post. Moonshot says "K3.1 Pro is capable of end-to-end delivery of complete financial workflows." That means work that ends in a model, a research note, or a decision, not a single answer.
Moonshot backs this with public and internal tests.
Public tests. Kimi K3.1 Pro "achieves state-of-the-art results on OfficeQA Pro v2." Moonshot says that test covers "complex fiscal data analysis over real financial documents through retrieval, cross-document verification, and multi-step numerical reasoning." The API documentation adds SpreadsheetBench v2. That test "evaluates end-to-end spreadsheet workflows including financial modeling, cross-sheet formula dependencies, and error diagnosis."
Frontier Finance. Moonshot says the model "also improves over K3 on Frontier Finance." That test "measures support for complete investment workflows from screening to profitability analysis."
Internal tests. Moonshot reports gains on three of its own finance tests:
| Internal test | What it checks, per Moonshot | Reported change |
|---|---|---|
| Financial Modeling | "formula linkage, financial logic, and final workbook quality" | Up 7.2 points over K3 |
| Financial Research | Whether models "can produce auditable research under evidence and cutoff constraints" | Up 11.6 points over K3 |
| Business Operation | "long-horizon management and decision-making in a simulated business environment" | About 4.2 times K3's score |
These are Moonshot's own numbers on its own tests. Treat them as a signal of focus, not as a guarantee for your files.
Factual accuracy. Moonshot writes that "Financial analysis and investment research demand a high degree of factual accuracy." It says it made "targeted improvements" to accuracy on real tasks. It also labels its accuracy test "a stress test," and says the rates "do not represent overall error rates in daily use."
In practice, this points to three kinds of finance work:
- Building a model. A three-statement or driver-based model in a spreadsheet, with linked formulas.
- Writing research. An investment note or company review, with each claim tied to evidence.
- Running analysis on documents. Pulling figures from filings, reports, and statements, then checking them across sources.
Whatever tool you use, keep a human check on the last step. Compare the key figures with the source documents before anything goes to a client or a board.
Coding, 3D, and safety changes
Finance is one of four focus areas in the launch post.
Coding. Moonshot says Kimi K3.1 Pro "covers the full range of coding work, from quick fixes in real codebases to full-project builds." It names game engines such as Unity, Unreal Engine, and Godot.
3D modeling. The post says the model "can write and run scripts for tools like Blender and CAD software." On BenchCAD, Moonshot reports 92.7% with tools.
Cybersecurity controls. Moonshot says that for the open-weight release, it "identified and removed cyber-related data from both training stages." It adds that the model's cyber capability "remains behind that of frontier proprietary models."
For business readers, the safety point matters most. An open-weight model can be run by anyone once the weights ship. Moonshot's post explains why it limited one risky skill before release.
How to access Kimi K3.1 Pro
Moonshot lists a separate path for each product.
| Where | How to start, per Moonshot |
|---|---|
| Kimi app and Kimi.com | Update the app on iOS, Android, or HarmonyOS, or visit kimi.com |
| Kimi Work | Download the desktop app, version 3.1.0 or later, for Windows or Apple silicon Macs |
| Kimi Code | Select Kimi K3.1 Pro with the /model command |
| Kimi API | Select kimi-k3.1-pro or kimi-k3.1-pro-highspeed |
| Kimi Business | Administrators can enable it; access is off by default at launch |
The API documentation lists three reasoning effort values: low, high, and max. Higher effort means deeper reasoning, more tokens, and more time.
For teams, Moonshot says "Zero Data Retention is available for API customers." That matters if your finance files are confidential.
API pricing
The Kimi API is billed per token. Its pricing page says model inference is "billed on a pay-as-you-go basis, with no subscription plan." Prices below are per million tokens.
| Model | Input | Output | Cached input |
|---|---|---|---|
| kimi-k3.1-pro | $2.00 | $10.00 | $0.20 |
| kimi-k3.1-pro-highspeed | $4.00 | $20.00 | $0.20 |
| kimi-k3 | $3.00 | $15.00 | $0.30 |
Two details stand out. First, Kimi K3.1 Pro costs less per token than Kimi K3 on this page. Second, the Highspeed variant doubles the base price for its extra speed.
The page says all K3 series models "support a 1,048,576-token context window." It also announces a change for older models. Starting October 31, 2026, K2 series models "will no longer be available to newly registered users." Moonshot plans to retire them by the end of December 2026.
The Kimi app uses separate membership plans. The API pricing page notes that token billing "differs from products such as the Kimi membership and Kimi Code."
What to watch next
A few open questions will shape how useful the release is over the coming weeks.
The weight release. Moonshot has promised the full weights by November 4. Once they ship, companies can run the model on their own servers. Check the license terms on release day before you plan a private deployment.
Independent tests on finance work. The finance results so far come from Moonshot. Outside testers will start running their own checks on spreadsheets and filings. Wait for a few of those before you move a reporting process.
The K2 retirement. New accounts lose K2 access on October 31, and all K2 models retire by the end of December. Teams on K2 should test the newer models before then.
Price changes. API prices often move after a launch. Re-read the pricing page before you commit to a monthly budget.
A simple test plan works for most teams. Pick one closed month of real data. Ask the model for a short variance note and a linked model. Then compare its figures, formulas, and claims with the work your team already signed off.
Kimi K3.1 Pro alternatives
Several other options cover the same finance and analysis work. The table uses each vendor's own pricing page, read on October 26, 2026.
| Option | What it offers | Price, per the vendor |
|---|---|---|
| Claude Opus 5.5 | Anthropic's model for "agentic coding and enterprise work" | API: $4 input and $20 output per million tokens |
| DeepSeek V4.1 Flash | An open model with a 1M context window | API: $0.30 input and $1.20 output per million tokens at peak; off-peak is half |
| Kimi K3 | The previous Kimi flagship | API: $3 input and $15 output per million tokens |
| Powerdrill Bloom | An AI workspace that turns finance files into models, reports, and dashboards | Free $0.00; Pro $13.27/mo |
Claude Opus 5.5 is a closed model. Anthropic's pricing page calls it a "Daily driver for agentic coding and enterprise work." In the Claude app, it comes with a paid Claude plan.
DeepSeek V4.1 Flash is the lowest-cost API option here. DeepSeek's pricing page says off-peak rates "are half of the peak rates." It also offers a 1M context window.
Kimi K3 stays available on the API. It suits teams that already tested their prompts on it, though it now costs more per token than the newer model.
Powerdrill Bloom is not a model. It is a workspace for running finance analysis on your own files. The next section explains where it fits.
Where Powerdrill Bloom fits
A model is one part of a finance workflow. The other parts are the files, the checks, and the finished output that people read.
Powerdrill Bloom is built around those parts. You upload statements, exports, or PDFs, then ask questions in natural language. Its homepage says "Every number comes back with the page, the row and the figure behind it."
Three tool pages match the finance work Moonshot describes:
- The AI Financial Modeling Tool builds "integrated three-statement financial models from your data and assumptions." It lists a sanity-check dashboard "with balance checks, ratios and red flags."
- The AI Financial Report Generator produces a monthly close report and a "Quarterly board financial deck with KPIs, segment view and asks."
- The AI Finance Dashboard Builder builds a CFO scorecard and a "Cash and working capital board with runway and AR/AP days."
Its Scheduled Agent covers recurring work. The changelog says "Every run finishes on schedule and delivers the results to your workspace, so the report is already waiting when you get there."
On price, the pricing page lists a Free plan at $0.00 and Pro at $13.27/mo. To compare it with a model you are testing, try Powerdrill Bloom with one set of statements you already know well.
Frequently asked questions
What is Kimi K3.1 Pro?
Kimi K3.1 Pro is a large language model from Moonshot AI, released on October 23, 2026. It is a post-trained version of Kimi K3 with a one-million-token context window. Moonshot highlights gains in coding, 3D creation, and financial research.
Is Kimi K3.1 Pro open source?
Moonshot calls it an open-weight model, but the weights were not public at launch. The launch post says the full model weights will be released by November 4, 2026. Until then, you can use it through Kimi's apps and API.
How much does the Kimi K3.1 Pro API cost?
Moonshot's pricing page lists $2.00 per million input tokens and $10.00 per million output tokens. Cached input costs $0.20 per million tokens. The Highspeed variant costs $4.00 for input and $20.00 for output.
Can Kimi K3.1 Pro build financial models?
Moonshot says the model can deliver complete financial workflows, including financial modeling. Its internal Financial Modeling test checks formula linkage and workbook quality, where it reports a 7.2-point gain over Kimi K3. Check any model against your source statements before use.
What are the alternatives to Kimi K3.1 Pro?
Alternatives include Claude Opus 5.5, DeepSeek V4.1 Flash, and the earlier Kimi K3. For finance reports and dashboards built from your own files, with each figure traced to its source, Powerdrill Bloom is another option.
Sources: Moonshot AI, Kimi K3.1 Pro · Kimi API Platform, Kimi K3.1 Pro · Kimi API Platform, Model pricing · Moonshot AI, Kimi K3 · Kimi on X, October 23, 2026 · Simon Willison, Kimi K3.1 Pro · Claude pricing · DeepSeek API pricing. Details read from vendor pages on October 26, 2026.