Kimi K3.1 Pro — Specifications
Compiled by Mustafa Ihsan from the vendor’s published documentation · Last updated
| Developer | Moonshot AI |
|---|---|
| Type | LLM (MoE, reasoning, multimodal) |
| Modality | Text, Image → Text |
| Parameters | 2.8T total / 104B active (MoE, 16 of 896 experts) |
| Context window | 1M |
| License | Kimi K3.1 Pro License (open weights) |
| Open weights | Yes |
| Released | 2026-10 |
| Input price | $2.00 /1M |
| Output price | $10.00 /1M |
| API providers | Moonshot (Kimi API), OpenRouter |
Run it locally
| VRAM (4-bit) | ~1.6 TB (native MXFP4 weights) |
|---|
Benchmarks
| Moonshot-reported | Terminal-Bench 4.0 65.7%; BenchCAD (with tools) 92.7%; OfficeQA Pro v2 64.4. Cached input $0.20 per 1M tokens; Highspeed variant $4.00 / $20.00. |
|---|
What is Kimi K3.1 Pro?
Kimi K3.1 Pro is Moonshot AI’s newest flagship, released on 24 October 2026. According to Moonshot, it is a post-trained version of Kimi K3 with substantially more post-training compute, aimed at coding, 3D creation, financial research and other agentic work. Like K3, it is a 2.8-trillion-parameter mixture-of-experts with 104B active parameters (16 of 896 experts per token) and a 1,048,576-token context window, and it reads text and images. It always thinks; reasoning effort can be set to low, high or max (the default). Moonshot reports 65.7% on Terminal-Bench 4.0, a vendor figure.
On Moonshot’s API, kimi-k3.1-pro costs $2.00 per million input tokens, $10.00 per million output tokens and $0.20 per million cached input tokens; writing to the cache costs $2.00 per million tokens for a 5-minute TTL or $4.00 for a 1-hour TTL. The Highspeed variant, which Moonshot says runs 2–3× faster, costs $4.00 in / $20.00 out. Batch jobs are billed at 60% of the standard price, and the model unlocks after a top-up of at least $1.
That is cheaper than its predecessor Kimi K3 ($3.00 / $15.00), and it is the model Moonshot’s retirement notice for the K2 series points to, which covers Kimi K2.7 Code. The weights are on Hugging Face under the Kimi K3.1 Pro License, which asks Model-as-a-Service businesses with more than $20 million in revenue over 12 months to sign a separate agreement. They ship as 4-bit MXFP4 files totalling about 1.56 TB, so self-hosting is a cluster-scale job and for most teams this is an API decision. Estimate a monthly bill with the AI API cost calculator.
Kimi K3.1 Pro pricing: API cost per 1M tokens
| Input (per 1M tokens) | $2.00 |
|---|---|
| Output (per 1M tokens) | $10.00 |
| Output/input ratio | 5× |
| Blended (4:1 in:out) | $3.60 per 1M tokens |
What Kimi K3.1 Pro costs per month
Real monthly spend at a 4:1 input-to-output mix — the ratio a typical chat or RAG workload actually produces.
| Workload | Tokens / month | Cost / month |
|---|---|---|
| Side project | 1M in / 0.25M out | $4.50 |
| Small team | 20M in / 5M out | $90 |
| Production | 200M in / 50M out | $900 |
Run your own numbers in the AI API cost calculator.
Cheaper alternatives to Kimi K3.1 Pro
| Model | Blended $/1M | You save |
|---|---|---|
| Llama 3.1 8B open | $0.0240 | 99% cheaper |
| Mistral NeMo 12B open | $0.0292 | 99% cheaper |
| Gemma 3 4B open | $0.0600 | 98% cheaper |
Self-host or pay the API?
Kimi K3.1 Pro is open-weight, so you can run it yourself. It needs ~1.6 TB (native MXFP4 weights) of VRAM at 4-bit. Self-hosting only beats the API once your volume is high enough to keep that hardware busy — the self-hosting vs API calculator works out the break-even point for your token volume.
Frequently asked questions
How much does Kimi K3.1 Pro cost per 1M tokens?
Kimi K3.1 Pro costs $2.00 per 1M input tokens and $10.00 per 1M output tokens. At a typical 4:1 input-to-output mix that blends to about $3.60 per 1M tokens.
How much does Kimi K3.1 Pro cost per month?
A small-team workload of 20M input and 5M output tokens a month costs about $90 on Kimi K3.1 Pro. A side project (1M in / 0.25M out) costs roughly $4.50.
What is a cheaper alternative to Kimi K3.1 Pro?
Llama 3.1 8B is the strongest cheaper option in our database at $0.0240 per 1M blended — about 99% less than Kimi K3.1 Pro. It is also open-weight, so self-hosting is an option.
Can I run Kimi K3.1 Pro locally?
Yes. Kimi K3.1 Pro is open-weight and needs about ~1.6 TB (native MXFP4 weights) of VRAM at 4-bit quantisation.
Why does Kimi K3.1 Pro charge more for output than input?
Output tokens are generated one at a time and cannot be batched the way a prompt can, so they cost the provider more to serve. Kimi K3.1 Pro charges 5× more for output, which is why prompt-heavy workloads are far cheaper to run than generation-heavy ones.
Prices are the published list rates for the model's primary API and are reviewed as providers change them. Volume, batch and cached-input discounts are not included. Compare every model side by side in the AI models database or the LLM leaderboard.