- Intelligence
- 12.7
- Rank 354 of 744 measured models
- Output speed
- 55.1 t/s
- Rank 302 of 403 measured models
- Price/1M
- $1.00
- Rank 385 of 659 measured models
- Context window
- 128K
- Rank 773 of 1199 measured models
Kimi K2 scores 12.7 on the Artificial Analysis Intelligence Index, ranking 354th of 744 measured models. The closest models on this index are GPT-4.1 (12.7), Granite 4.2 30B (12.7), and Qwen3 235B A22B 2507 (Reasoning) (12.7). At $1.00 per million tokens (3:1 input/output blended price), it is cheaper than GPT-4.1 ($3.50), more expensive than Granite 4.2 30B ($0.282) and Qwen3 235B A22B 2507 (Reasoning) ($0.747). With an output speed of 55.1 t/s, it is slower than GPT-4.1 (152 t/s), Granite 4.2 30B (75.5 t/s), and Qwen3 235B A22B 2507 (Reasoning) (85.4 t/s).
Additional benchmarks5
Design Arena · Codecategories51.7%
Design Arena · Dataviz49.4%
Design Arena · Gamedev46.4%
Design Arena · Uicomponent55.1%
Design Arena · Website53.1%
Models at a similar level
The closest models on the Intelligence Index, one configuration per family.
- GPT-4.1
OpenAI · $3.50 · 152 t/s
12.7$3.50152 t/sCompare - Granite 4.2 30B
IBM · $0.282 · 75.5 t/s
12.7$0.28275.5 t/sCompare - Qwen3 235B A22B 2507 (Reasoning)
Alibaba · $0.747 · 85.4 t/s
12.7$0.74785.4 t/sCompare - Qwen3 235B A22B Thinking 2507
Qwen · $1.21 · —
12.7$1.21—Compare
