Models / Kimi K3.1 Pro
Kimi K3.1 Pro
Kimi K3.1 series family · Provider: moonshot-ai · Status: active
Kimi K3.1 Pro is a Kimi K3.1 series model developed by moonshot-ai, a Chinese AI company headquartered in 13F, Building 1, JD Technology Building, 76 Zhichun Road, Haidian District, Beijing, China with a 1,048,576-token context window and open weights.
| Key fact | Value |
|---|---|
| Model ID | kimi-k3.1-pro |
| Architecture | Post-trained from Kimi K3 with substantially larger post-training compute; provider describes it as a 3T-class open-weight model |
| Context window | 1,048,576 tokens |
| Open weights | Yes |
| License | Unknown |
| Self-hosting | Yes |
| API available | Yes |
| API pricing | ¥13 input / ¥68 output per 1M tokens (CNY) · provider pricing page |
| Cloud providers | Moonshot AI Platform |
Capabilities
| Capability | Supported |
|---|---|
| Reasoning | Yes |
| Coding | Yes |
| Math | Unknown |
| Chinese | Unknown |
| English | Unknown |
| Multilingual | Unknown |
| Vision | Yes |
| Audio | Unknown |
| Video | Unknown |
| Tool calling | Yes |
| Function calling | Unknown |
| Structured output | Yes |
| Agent capability | Yes |
| RAG | Unknown |
| Computer use | Unknown |
Known limitations
- Priced in CNY on the China platform; the international USD list was not published when this record was created, so cross-provider comparison needs an explicit exchange-rate assumption
- Cache writes are billed separately by TTL tier (5 minutes / 1 hour); partial blocks that do not fill a cache block are billed as cache misses
Kimi K3.1 Pro is Moonshot’s first K3.1-series model: a post-trained version of Kimi K3 with substantially larger post-training compute, a 1,048,576-token context window, vision input and reasoning effort control. The provider describes it as its most capable 3T-class open-weight model.
Pricing
The China platform lists ¥13 per 1M input tokens, ¥68 per 1M output tokens, with cached input at ¥1.3 per 1M — a 90% cache discount. A kimi-k3.1-pro-highspeed variant lists ¥26 / ¥136 with roughly 2–3x the throughput.
At an indicative 7.1 CNY/USD, that is about $1.83 input / $9.58 output per 1M tokens. The comparison to the previous flagship is the point: Kimi K3 lists $3.00/$15.00 on the international platform, so K3.1 Pro is roughly a third cheaper on input and a third cheaper on output — consistent with the provider’s own “about one-third lower” framing.
The number to be careful with is the currency. This record is in CNY because that is how the price was published; the international USD list was not available when the record was created. Any cross-provider comparison that mixes this row with USD rows without stating an exchange rate is not a comparison.
What changed from K3
- Larger post-training, described by the provider as joint RL across hundreds of downstream domains, rather than a new base model.
- Flat pricing: no tiering by context length, but cache writes are billed separately by TTL (5 minutes or 1 hour).
- Batch inference at 60% of the standard price, for
kimi-k3.1-proandkimi-k3. - A highspeed variant at exactly double the standard price.
How to read it
China AI Hub analysis: K3.1 Pro is a price cut wrapped in a capability release. The architecture is inherited from K3; the changes are post-training and packaging. That makes the migration decision unusually simple for existing K3 users — same window, same deployment model, lower price — and unusually hard for K2 users, whose models retire by the end of December 2026 and who now face a flagship price instead of a mid-tier one.
Verify the CNY price on the platform before budgeting: it is the number most likely to move, and the international list may appear at a different level.
Provider
Pricing
Related models (same family)
API
Model family timeline
| Model | Released | Status |
|---|---|---|
| Kimi K3.1 Pro Highspeed | — | active |
| Kimi K3.1 Pro (this page) | — | active |
Data interpretation
| Field | Value | Evidence type |
|---|---|---|
| Context window | 1,048,576 tokens | Official |
| Architecture | Post-trained from Kimi K3 with substantially larger post-training compute; provider describes it as a 3T-class open-weight model | Official |
| Open weights | Yes | Official |
| API pricing | ¥13 / ¥68 per 1M tokens (CNY) | Official |
| Family position | 2 models in the Kimi K3.1 series family | China AI Hub analysis |
Evidence types: Official = vendor documentation, pricing pages or model cards. Vendor-reported = benchmark scores published by the vendor. China AI Hub analysis = derived from the database itself. See the sourcing policy.
Sources
Confidence and source hierarchy per the sourcing policy. Facts change; check the source before relying on this page.
What is the context window of Kimi K3.1 Pro?
Kimi K3.1 Pro has a 1,048,576-token context window.
Is Kimi K3.1 Pro open weight?
Yes — Kimi K3.1 Pro weights are openly available.
How much does Kimi K3.1 Pro cost through the API?
¥13 per 1M input tokens and ¥68 per 1M output tokens (CNY).
Where does China AI Hub get its Kimi K3.1 Pro data?
From 2 sources (official pages first), last verified 2026-11-05. Benchmark scores are labeled by source type; see the sourcing policy for details.