ModelPickr

Compare / Kimi

Kimi · canonical pair

Kimi K2.6 vs Kimi K3

Alphabetical by model id, so this URL is the one to index. Kimi K2.6 is moonshotai/kimi-k2.6. Kimi K3 is moonshotai/kimi-k3. Both are among the models the most providers list, inside the Kimi group.

A

Kimi K2.6

$11.70

per month at the reference workload

  • Reasoning
  • Tool use
  • Structured output
  • Vision
  • Long context
  • Open weights

vs

B

Kimi K3

$40.50

per month at the reference workload

  • Reasoning
  • Tool use
  • Structured output
  • Vision
  • Long context
  • Open weights
100 requests a day, 2,000 input tokens and 500 output tokens, no cache hits. A light chat workload so the rate cards are comparable.
SpecKimi K2.6Kimi K3
Monthly, reference workload$11.70$40.50
Input / 1M$0.95$3.00
Output / 1M$4.00$15.00
Cache read / 1M$0.16$0.30
Cache write / 1M—$3.00
Context262,1441,048,576
Max output262,144131,072
ReasoningYesYes
Tool callsYesYes
VisionYesYes
Open weightsYesYes
Providers5280
Released2026-04-212026-07-16

Same workload, your numbers

Prices are USD per 1M tokens. A month is 30 days. Cache hits use the cache-read rate when the model has one; otherwise the cache field is ignored.

Estimated monthly cost

Kimi K3

$40.50

Kimi K3 costs $28.80 more per month than Kimi K2.6 at these inputs.

monthly = requests/day × 30 × (input tokens × (1 − cache hit) × input price + input tokens × cache hit × cache-read price + output tokens × output price) ÷ 1,000,000