Moonshot AI
Kimi K3.1 Pro
The 3T-class challenger that undercuts Opus
What it is
Moonshot’s flagship: a 3T-class open-weight model with a 1M context and native visual understanding, positioned for long-horizon coding and knowledge work. The weights are promised by November 4, 2026; the API lists at $2/$10 per 1M tokens.
NLT take
The most interesting open-weight challenge to Opus on price: $2/$10 against Opus’s $4/$20, with a 1M window. The caveat is the weights, not the rate.
Before you switch
No release date is published in the docs we could read, and the weights land after launch. That is why the card has no Hugging Face link yet.
Cost for a real task
The same task on every model: 50,000 input tokens and 10,000 output tokens, at the base tier.
Task cost
$0.20
50K input + 10K output tokens
- Input, per 1M
- $2.00
- Output, per 1M
- $10.00
- Cached input, per 1M
- $0.20
Pricing source: OpenRouter listing
Specs
- Maker
- Moonshot AI
- Licence
- Open weights (announced)
- Released
- October 23, 2026
- Context window
- 1,000,000 tokens
- Parameters
- 3T-class
- Input / output
- Text · Image → Text
Also known as: kimi-k3.1-pro · kimi-k3.1-pro-highspeed
Watch it in action
Hands-on videos from creators we trust. They open on YouTube; we do not host or re-upload them.
Kimi K3.1 Pro Is INSANE! CHEAPER & FASTER Than Claude Opus 5.5?!WorldofAIWhy watch: Tests the cheaper-than-Opus claim, which is the whole pitch.Watch on YouTube(opens YouTube in a new tab)
Kimi K3.1 Pro Is HERE – Does THIS Really Beat Claude & GPT?Bijan BowenWhy watch: A second opinion on whether it beats the frontier or only undercuts it.Watch on YouTube(opens YouTube in a new tab)
Kimi K3.1 Is the #1 Open Model and It Beat GPT-6 Astra at Coding!Universe of AIWhy watch: The strongest claim in the launch cycle, tested against Astra.Watch on YouTube(opens YouTube in a new tab)
From our coverage
Articles about this model or its family. A “Background” tag means we have not covered this exact release yet.
- BackgroundJuly 22, 2026Kimi K3 vs Fable 5, GPT-5.6, Grok: 2026 BenchmarksKimi K3 is the first open 3T-class model. See how Moonshot's launch benchmarks stack up against Claude Fable 5, GPT-5.6 Sol, Grok 4.5, and Gemini in 2026.
- BackgroundApril 27, 2026Kimi K2.6: Open-Weight 300-Agent Swarm Tops GPT-5.4 (2026)Moonshot AI's Kimi K2.6 (April 20, 2026): a 1T-parameter open-weight MoE that orchestrates 300 sub-agents and tops GPT-5.4 on SWE-Bench Pro at 58.6%.
Sources & model links
Every number on this card traces back to one of these links. The maker page proves the positioning, the pricing page proves the rates, the Hugging Face repo proves the weights exist, and the Ollama page proves a local tag.
Official
- Moonshot AIVerifies: maker
- OpenRouter listing ($2/$10)Verifies: priceMoonshot’s own pricing table is rendered client-side; the USD rate here is the published OpenRouter listing.
- Kimi K3.1 Pro documentationVerifies: specs · licenceOpen weights are promised by November 4, 2026; no public Hugging Face repo at launch.
Run locally
- Run with Ollama (K3, not K3.1)Verifies: local runOllama carries Kimi K3 under the cloud tag; K3.1 Pro is not published there yet.