Primary use stated or inferred cautiously from official documentation.
K
MOONSHOT AI / moonshotai/Kimi-K3.1-Pro · API kimi-k3.1-pro
Kimi K3.1 Pro
Moonshot AI MoE model with 2.8 trillion total parameters and 104 billion activated per token. The marketing term 3T-class is not a different exact count. Its API launched in October and 96 downloadable weight shards have now been verified. The profile pins the checkpoint and its own license.
01 / SUMMARY
The essential
Maximum input capacity when published by the source.
Documented maximum generation limit.
Verified availability channels.
Primary functional family and verified specialties.
Declared terms for API access, model weights, or self-hosting.
02 / TECHNICAL SHEET
Limits and integration
| API ID | moonshotai/Kimi-K3.1-Pro · API kimi-k3.1-pro |
|---|---|
| Model type | Multimodal · Code · Text and reasoning |
| Access model | Open weights |
| License | Kimi K3.1-Pro License |
| Deployment | Self-hosted |
| Release | 24 OCT 2026 |
| Knowledge cutoff | Not published |
| Entrance | Text · Image |
| Exit | Text · Code |
| Context window | 1.048.576 tokens |
| maximum output | Not published |
| Reasoning | Not published |
| Published tools | |
| Structured Outings | Not published |
| Batch processing | Not published |
| Prompt cache | Not published |
| Fine-tuning | Not published |
| Verified platforms | Hugging Face · PyTorch |
Research and evaluation with your own data, preserving the exact weight revision, configuration and dependencies. Compare accuracy, memory and latency before integrating the model into an application.
The license permits uses subject to conditions and is not equivalent to MIT or Apache 2.0. It includes particular obligations for Model as a Service businesses and certain large-scale products. Weights use MXFP4 quantization with MXFP8 activations and require specialized infrastructure. API rates and maximum output are not inferred from checkpoint size.
03 / CAPABILITIES
What it can do
Orientation
Research and evaluation with your own data, preserving the exact weight revision, configuration and dependencies. Compare accuracy, memory and latency before integrating the model into an application.
Context
1.048.576 tokens · Not published
Tools and integration
Not published
Access
Downloadable weights and Kimi API
Input modalities
TextImage
04 / PRICES
Documented cost
| Concept | Worth | Unit/condition |
|---|---|---|
| Standard input | See the API channel’s official pricing | Standard API rate |
| Cached input | Not published | Reading reused prefixes |
| Cache write or storage | Not published | The condition varies by provider |
| Standard output | See the API channel’s official pricing | May include reasoning tokens |
| Batch input | Not published | Asynchronous processing |
| Batch output | Not published | Asynchronous processing |
This profile covers the downloadable checkpoint and retains a link to the official terms. API, Batch and cache rates must be checked for the channel used. Deploying 2.8 trillion parameters depends on runtime, quantization and distribution; downloadable weights do not mean free inference.
i Prices change and may depend on level, region or context length. Check the source before making a decision.
05 / EVALUATIONS
How to read the results
“
The card reports 65.7 on Terminal-Bench 4.0 with Claude Code and 94.2 on GPQA-Diamond. Unless noted otherwise it uses max effort, temperature 1.0 and top_p 0.95. These are Moonshot results not reproduced by Inferama. Each test’s notes are preserved and figures from different protocols or judges are not equated.
| Benchmark | Result | Metric | Source |
|---|---|---|---|
| Terminal-Bench 4.0 | 65,7 / 100 | Score | View source ↗ |
| GPQA Diamond | 94,2 / 100 | Score | View source ↗ |
Inferama only highlights a “best result” when the metric, test set, configuration, and date allow for an equivalent comparison. The supplier's figures are presented as claims from its own source.
06 / VERSIONS
Chronology
Kimi K3.1 Pro · pesos verificados
Verified checkpoint availability; does not replace the API release date.
Kimi K3.1 Pro · API
API listing recorded in TokenHub’s official history.
07 / SOURCES