LLM.co · Private AI economicsKimi K3.1 Pro Hosting Cost Calculator
Moonshot's API or your own GPUs? Price Kimi K3.1 Pro both ways, with verified GPU rates and the memory its 4-bit, multi-terabyte checkpoint really needs.
Open weights: compared against API pricing and self-hosting on B300 (Blackwell Ultra).
What Kimi K3.1 Pro costs to run
Kimi K3.1 Pro is Moonshot AI's current flagship: 2,800B parameters, 104B active per token, a 1.05M-token context window and weights trained natively in MXFP4. Moonshot sells it at $2.00 input, $0.20 cached and $10.00 output per million tokens, a third below its predecessor Kimi K3 ($3.00 / $15.00). On a workload of 2B input and 400M output tokens a month with half the input cached, that's $6,200 a month on K3.1 Pro against $9,300 on K3.
Hardware that can hold it
The 1,561 GB checkpoint needs about 1,795 GB with runtime headroom, and about 2,342 GB to batch many users. Because the weights are already 4-bit, there is no smaller official build to fall back on. GPUs with native FP4 math (Blackwell, MI355X) run it at full speed. On Hopper, the 4-bit weights load through dequantizing kernels.
| GPU | GPUs to fit | Topology | Cheapest verified on-demand | Monthly at 730 h |
|---|---|---|---|---|
| NVIDIA B300 (Blackwell Ultra)288 GB · FP4 native | 8 | One node | $6.94 · RunPod | $40,530 |
| AMD Instinct MI355X288 GB · FP4 native | 8 | One node | $8.60 · Oracle Cloud (OCI) | $50,224 |
| NVIDIA H200 141GB141 GB · FP4 via dequantization | 16 | 2 nodes: needs multi-node serving | $3.59 · RunPod | $41,931 |
| NVIDIA B200180 GB · FP4 native | 16 | 2 nodes: needs multi-node serving | $5.98 · RunPod | $69,846 |
License
Kimi K3.1-Pro License (custom, MIT-style): Modified-MIT style. Free commercial use, modification, redistribution. (a) If licensee/affiliates operate a Model-as-a-Service business with >US$20M revenue over any 12 months, need a separate agreement with Moonshot for commercial use. (b) Products with >100M MAU or >US$20M monthly revenue must prominently display the model name in the UI. Internal use and use via Moonshot official products/certified inference partners are exempt from (a)/(b). Read the license text before you resell inference.
The calculators below open on Kimi K3.1 Pro on 8× B300. The worked break-even on the main calculator page compares every B300 and B200 offering in the registry.
| Model | Params (total / active) | Context | Weights | License | API $/1M in · cached · out | Source |
|---|---|---|---|---|---|---|
| Kimi K3.1 ProMoonshot AI · 2026-10 | 2,800B / 104B | 1,049K | 1,561 GBMXFP4 weights (quantization-aware training) | Kimi K3.1-Pro License (custom, MIT-style)Modified-MIT style. Free commercial use, modification, redistribution. (a) If licensee/affiliates operate a Model-as-a-Service business with >US$20M revenue ove… | $2.00 · $0.20 · $10.00Moonshot (official) | Verified platform.kimi.ai |
| Kimi K3Moonshot AI · 2026-06 | 2,800B / 104B | 1,049K | 1,561 GBMXFP4 weights (quantization-aware training) | Kimi K3 License (custom, MIT-style; same terms as K3.1-Pro with model name 'Kimi K3')Modified-MIT style. Free commercial use, modification, redistribution. (a) If licensee/affiliates operate a Model-as-a-Service business with >US$20M revenue ove… | $3.00 · $0.30 · $15.00Moonshot (official) | Verified platform.kimi.ai |
| GPU | Provider · offering | $/GPU-hr | As listed | 730 h / GPU | Source |
|---|---|---|---|---|---|
| B300 (Blackwell Ultra)B300 | RunPodCommunity Cloud | $6.94 | $6.94per GPU | $5,066 | Verified runpod.io |
| B300 (Blackwell Ultra)B300 | Hyperstackon-demand · Hyperstack VM (B300) | $7.40 | $7.40per GPU | $5,402 | Verified hyperstack.cloud |
| B300 (Blackwell Ultra)B300 SXM (B300-SXM-8-288G) | ScalewayOn-demand 8-GPU instance · B300-SXM-8-288G | $8.29 | $66.32per 8-GPU instance | $6,052 | Unverified computeprices.com |
| B300 (Blackwell Ultra)B300 (Blackwell Ultra) | Crusoeon-demand (current) · Crusoe GPU instance | $8.96 | $8.96per GPU | $6,541 | Verified crusoe.ai |
| B300 (Blackwell Ultra)B300 | RunPodSecure Cloud on-demand | $8.99 | $8.99per GPU | $6,563 | Verified runpod.io |
| B300 (Blackwell Ultra)HGX B300 | NebiusOn-demand | $9.50 | $9.50per GPU, 8-GPU nodes | $6,935 | Verified nebius.com |
| B300 (Blackwell Ultra)B300 | RunPodInstant Cluster (Clusters product) | $10.79 | $10.79per GPU, 8-GPU nodes | $7,877 | Verified runpod.io |
| B300 (Blackwell Ultra)B300 SXM6 268GB | VerdaGPU instance on-demand | $12.41 | $12.41per GPU | $9,059 | Verified verda.com |
| B300 (Blackwell Ultra)B300 SXM6 268GB | VerdaInstant cluster (16-144 GPUs) | $12.41 | $12.41per GPU, 16-GPU nodes | $9,059 | Verified verda.com |
| B300 (Blackwell Ultra)B300 (Blackwell Ultra) | Oracle Cloud (OCI)on-demand (pay as you go) · B300 bare metal (shape name unconfirmed) | $15.00 | $15.00per GPU | $10,950 | Verified oracle.com |
| B300 (Blackwell Ultra)B300 (Blackwell Ultra) | AWSon-demand · p6-b300.48xlarge | $17.80 | $142.42per 8-GPU instance | $12,995 | Verified aws.amazon.com |
| Instinct MI355XAMD MI355X | Oracle Cloud (OCI)on-demand (pay as you go) · MI355X shape (name unconfirmed) | $8.60 | $8.60per GPU | $6,278 | Verified oracle.com |
| H200 141GBH200 SXM | Fluidstackon-demand · Fluidstack H200 SXM | $2.30 | $2.30per GPU | $1,679 | Unverified gpuperhour.com |
| H200 141GBH200 SXM | Hyperstackreserved ('starting from') · Hyperstack VM (H200 SXM) · term not stated | $2.79 | $2.79per GPU | $2,037 | Verified hyperstack.cloud |
| H200 141GBH200 | RunPodCommunity Cloud | $3.59 | $3.59per GPU | $2,621 | Verified runpod.io |
| H200 141GBH200 SXM | Hyperstackon-demand · Hyperstack VM (H200 SXM) | $3.99 | $3.99per GPU | $2,913 | Verified hyperstack.cloud |
| H200 141GBH200 | Crusoeon-demand (current) · Crusoe GPU instance | $4.29 | $4.29per GPU | $3,132 | Verified crusoe.ai |
| H200 141GBH200 | RunPodSecure Cloud on-demand | $5.29 | $5.29per GPU | $3,862 | Verified runpod.io |
| H200 141GBHGX H200 | NebiusOn-demand | $5.40 | $5.40per GPU, 8-GPU nodes | $3,942 | Verified nebius.com |
| H200 141GBHGX H200 | CoreWeaveOn-demand 8-GPU node · HGX H200 | $6.30 | $50.44per 8-GPU instance | $4,603 | Verified coreweave.com |
| H200 141GBH200 SXM | RunPodInstant Cluster (Clusters product) | $6.35 | $6.35per GPU | $4,636 | Verified runpod.io |
| H200 141GBH200 SXM5 141GB | VerdaInstant cluster (16-144 GPUs) | $6.70 | $6.70per GPU, 16-GPU nodes | $4,891 | Verified verda.com |
| H200 141GBH200 SXM5 141GB | VerdaGPU instance on-demand | $6.70 | $6.70per GPU | $4,892 | Verified verda.com |
| H200 141GBH200 | AWSon-demand · p5en.48xlarge | $7.91 | $63.30per 8-GPU instance | $5,776 | Verified aws.amazon.com |
| H200 141GBH200 | Oracle Cloud (OCI)on-demand (pay as you go) · BM.GPU.H200.8 | $10.00 | $10.00per GPU | $7,300 | Verified oracle.com |
| H200 141GBH200 | Google Cloudon-demand · a3-ultragpu-8g | $10.60 | $84.81per 8-GPU instance | $7,739 | Verified cloud.google.com |
Registry last checked 2026-11-14. Prices older than 45 days are flagged stale. Verified means read from the provider's own page on that date; unverified means the official page couldn't be read and the figure came from a third-party listing (named in the link); estimatemeans a reseller or analyst figure. A missing price is shown as unavailable, never filled in. Dense TFLOPS are half the sparse figures vendors headline. “n/p” means not published.
API token cost calculator
Estimate pay-as-you-go inference spend from users and requests, or from monthly token totals. Cached input, cache writes and reasoning tokens are each priced separately and never double-counted.
Reasoning tokens bill at the output rate and are added to output below, once.
220K tasks → 660K model calls a month. Context overhead is 60% of input tokens.
Dedicated GPU hosting cost calculator
Rent GPUs by the hour and serve an open-weight model yourself. Compare providers for the same configuration, and see whether the model fits, how fast it might run and what each token costs at your utilization.
2,800B total · 104B active · ships as MXFP4 weights (quantization-aware training) · weights ≈ 1,561 GB at FP4 (MXFP4/NVFP4)
Sold as whole nodes means you pay for every GPU in the node, even if your replica needs fewer. Same storage, network and management assumptions applied to every row. Throughput is a planning estimate (memory-bandwidth roofline, 60% efficiency). Load-test before you commit.
Private LLM hardware ownership calculator
Buy the servers and run them in your facility or a colocation cage. Cash flow, accounting depreciation and economic cost are shown separately, because they answer different questions.
| View | Monthly | What it means |
|---|---|---|
| Cash flow | $849,800 | Capex paid up front, then opex each month |
| Depreciation | $15,205 | Straight-line (cost − residual) over 4 years: the P&L charge |
| Economic | $18,617 | Level capital cost including the 8% cost of money: used for comparisons |
We size, procure, install and operate private LLM hardware, and hand you the keys.
API vs. self-hosting break-even calculator
The four strategies on one chart. Slide the volume and watch where fixed infrastructure overtakes per-token pricing. Infrastructure adds replicas as volume grows, and a line breaks where a strategy runs out of capacity.
| Strategy | Monthly | Annual | 3-yr | 5-yr |
|---|---|---|---|---|
| API inferencepay per token | $31,000 | $372K | $2.14M | $5.95M |
| Dedicated cloud GPUs1 unit of 8 GPUs | $61,407 | $737K | $4.3M | $10.7M |
| Owned hardware1 unit of 16 GPUs | $48,946 | $587K | $2.1M | $4.94M |
| Hybrid30% API, 1 owned unit | $58,246 | $699K | $2.4M | $5.7M |
- 1 Owned hardware matches the API at 15.8B input tokens/mo ($48,946/mo)
- 2 Hybrid matches the API at 22.6B input tokens/mo ($69,923/mo)
Financial break-even is not proof of capacity. It's the volume where a fixed monthly bill equals the API bill. Whether one replica can serve that volume depends on throughput, so the chart adds units as volume grows and breaks the line where a strategy runs out of room.
Agentic AI infrastructure cost calculator
Agents call models many times per task, resend context at every step and retry when tools fail. Start from a preset, then route each share of traffic to a premium API, a low-cost API or a private model.
Long repository context, many tool calls per change, heavy reasoning. Illustrative starting values, not benchmarks.
Agent workloads resend context on every step, so input tokens dominate: 9.11B input vs 524M output here. Caching and routing usually save more than haggling over output prices.
Enterprise deployment recommendation
A deterministic recommendation from your calculator inputs plus the factors cost can't capture: data sensitivity, regulation, team capability and licensing.
Also uses: 10B input tokens/mo, 50% growth, Kimi K3.1 Pro licensing, and the infrastructure from calculators 2–3.
API-first deployment
Use pay-as-you-go APIs now; revisit when volume or data rules change.
- At this volume the API is the lowest-cost option ($31,000/month vs $48,946 self-hosted).
- Pay-per-token has no idle capacity, no capex and nothing to operate.
The rules are fixed and run in your browser, so the same inputs always give the same answer. Cost is one input among several: the cheapest option loses if it fails on data handling, team capability, licensing or capacity.
Stop renting intelligence. Start owning your AI infrastructure.
An engineer reviews your scenario against real quotes, your traffic logs and your data rules, then returns a costed architecture: API, dedicated, owned or hybrid.
Get a custom enterprise LLM infrastructure assessment.
Ask for an infrastructure assessment, a private LLM deployment quote or a consultation. You decide whether your calculator scenario is attached. We use it only to answer you.
LLM hosting cost questions.
Is Kimi K3.1 Pro open source?
Its weights are open under a custom MIT-style license. Commercial use is allowed. Model-as-a-Service businesses above US$20M in revenue need a separate agreement, and very large products must display the model name.
How many GPUs does Kimi K3.1 Pro need?
About 1.8 TB of GPU memory before KV cache. That is one 8× B300 or 8× MI355X node, or two 8-GPU nodes of B200 or H200, which then need multi-node serving.
Is it cheaper to self-host Kimi K3.1 Pro than to use Moonshot's API?
Only at very high, steady volume. A dedicated 8× B300 node costs tens of thousands of dollars a month, so at the API's prices it breaks even in the tens of billions of input tokens per month, if the node can serve that much. Use the break-even calculator with your own ratio and cache rate.
Which inference engines support Kimi K3.1 Pro?
The model card documents vLLM, SGLang and TokenSpeed.