Kimi K3.1 Pro
Kimi K3.1 Pro is Moonshot AI's most capable model to date, released October 24, 2026. 3-trillion-class parameters with native visual understanding and a 1M-token context window. Priced at CNY 13/68 per 1M tokens for input/output with cache hits at CNY 1.30 (10% of input). Terminal-Bench 4.0 scores 65.7%, ahead of GPT-6 Astra (58.2%) and close to Claude Opus 5.5 (66.4%). Open weights are due before November 4, 2026.
Best For
Strengths
- 1M-token context window
- 3T-class parameters with native vision
- Cache hits at only 10% of input price
- Terminal-Bench 4.0: 65.7%
- Highspeed variant at 2-3x output speed
- Open weights before Nov 4, 2026
Weaknesses
- Higher output price than most domestic peers
- Newer ecosystem than Kimi K3
- Intl platform pricing not yet listed on the CN portal
Related Alternatives
Kimi K3
Same 1M context, 35% higher price
Claude Opus 5.5
Terminal-Bench 66.4%, pricier per token
Frequently Asked Questions
How much does Kimi K3.1 Pro cost?
CNY 13 per 1M input tokens and CNY 68 per 1M output tokens, with cache hits at CNY 1.30 per 1M (10% of the input price). Context cache writes are billed separately by TTL: CNY 13 for the 5-minute tier and CNY 26 for the 1-hour tier. The context window is 1,048,576 tokens and billing does not tier by context length.
What is the difference between Kimi K3.1 Pro and Kimi K3?
Both support a 1,048,576-token context window, but K3.1 Pro is 35% cheaper on both input (CNY 13 vs CNY 20) and output (CNY 68 vs CNY 100) while scoring higher on Terminal-Bench 4.0 (65.7% vs K3's 55.95% on Kimi Code Bench 3.0 Long-horizon). K3.1 Pro also adds native visual understanding and a Highspeed variant at 2-3x output speed.
See Kimi K3.1 Pro's full record in our database
Live pricing for all Moonshot AI plans, token cost calculator, and the latest price changes.
View Full Comparison →