None
None
Long Story Video Skill
Turn a brief into a 10s–10min piece — the agent drafts the script, prompts and shots in one pass.
Ads Video Skill
Build product ads end to end, with scripts, prompts and footage generated together.
3D Science Explainer Video Skill
Turn scientific concepts into 3D explainer animations
Feedback
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
kimi k3.1 pro
Run coding, CAD and research tasks through Kimi K3.1 Pro — an open-weight 3T-class model with a 1M-token window and a 2-3x faster mode.
All Tools
Discover our comprehensive AI-powered animation toolkit
MiniMax H3
MiniMax H3 AI Video Generator
Seedance 2.5
The Future of AI Video Is Here.
Seedance 2.0
The Future of AI Video Is Here.
Veo3.1
Create Stunning Videos with Veo3.1
FLUX 3 Video Generator
Kling 3.0
Next-Gen AI Video Generator
Grok Video Generator
Create Videos from Text or Images with AI
MiniMax H3 video generator
MiniMax H3 AI Video Generator
Where Kimi K3.1 Pro Earns Its Place
An open-weight release built at 3T-class scale, carrying a one-million-token window and a Highspeed option for faster turns.
- Open Weights at 3T Scale, a Million-Token WindowBuilt as the debut of the K3.1 line, this release pushes open-weight capability to 3T scale and keeps a full million tokens in view for code, 3D assets and finance.
- A Highspeed Mode That Answers 2-3x FasterSwitch to the Highspeed variant when you need quicker back-and-forth; it returns replies two to three times faster than the standard build, and both are shipping now.
- Reinforcement Learning Tuned Across Hundreds of FieldsPost-training was rebuilt end to end — RL framework, long-context inference, sandboxes and rollout scheduling — on far more compute, so the gains hold up across domains.
Three Ways to Reach Kimi K3.1 Pro
Pick the entry point that fits your team: the web and mobile apps, the desktop coding tools, or the API and enterprise controls.
Capabilities You Get with Kimi K3.1 Pro
From end-to-end coding and parametric CAD to financial research, hallucination reduction and benchmark results — here is where the model performs.
Full-Project Coding Across Game Engines
It handles small patches in live repositories as well as complete builds, and works inside Unity, Unreal Engine and Godot with tighter typography, motion and front-end finish.
Self-Debugging on Long-Running Missions
In an internal walking-simulator run spanning 167 missions and 420 reviews, the model traced failures and fixed 36 shoulder moment-arm errors before retraining its controller.
Parametric CAD and 3D Modeling
It posts 92.7% on BenchCAD with tools and writes Blender and CAD scripts straight through — no MCP connectors — for parts, buildings and custom furniture.
Auditable Research and Finance Workflows
A 64.4 on OfficeQA Pro v2, with gains over K3 of 7.2 points in financial modeling, 11.6 points in research and roughly 4.2x in simulated business operation.
Fewer Hallucinations on Real Work
Focused training pulled claim-level hallucination on Kimi Hallu Bench Hard from 23.4% down to 8.3%, and answers carrying at least one factual error fell to 42.4% of the earlier rate.
Benchmark Results and Safer Open Weights
The model lands 65.7% on Terminal-Bench 4.0, ahead of GPT-6 Astra at 58.2%, while cyber-related data removal moved xdev-bench-hard from 90.5% to 81.0%.
Kimi K3.1 Pro: Questions Answered
Straight answers on context length, Highspeed differences, weight release timing, benchmarks and enterprise access.
How much context can Kimi K3.1 Pro hold at once?
It carries a one-million-token window. That makes it both the opening model of the K3.1 line and the strongest 3T-class open-weight release available so far.
How does the Highspeed variant differ from the standard one?
Highspeed responds two to three times faster than the standard build, which suits rapid iteration. Both are live now across Kimi.ai, Kimi Work, Kimi Code and the Kimi API.
When do the full model weights land?
Full weights are scheduled for release by November 4, 2026. The open-weight rollout follows published principles, including keeping offensive cyber capability below the frontier.
Where does it stand against proprietary frontier models?
On coding, agentic and knowledge benchmarks it matches or nears the strongest closed models: 65.7% on Terminal-Bench 4.0 against 58.2% for GPT-6 Astra and 66.4% for Claude Opus 5.5, plus 92.7% on BenchCAD with tools.
Can finance teams rely on it for research?
It scores 64.4 on OfficeQA Pro v2 and beats K3 in modeling and research, while claim-level hallucination on Kimi Hallu Bench Hard dropped from 23.4% to 8.3% — a deliberate stress test rather than a daily error rate.
How do enterprise teams switch it on?
Kimi Enterprise admins can enable it for their organization, since access is off by default at launch. Kimi Business adds enterprise-grade privacy, member management and split personal and org accounts.
Start Building with Kimi K3.1 Pro
Kimi K3.1 Pro is available now across Kimi.ai, the mobile apps, Kimi Work, Kimi Code and the Kimi API. Admins can enable it in Kimi Enterprise, or teams can subscribe to Kimi Business.