Kimi K2 price vs K3.1 Pro: what the K2 retirement costs you
Updated 2026-10-25. Prices from Moonshot's official pricing page, via models.json.
Moonshot is retiring its Kimi K2 series. If you run agents on K2.6 or K2.7 Code, the natural move is Kimi K3.1 Pro, and it costs more per token. Here is how much, using Moonshot's official prices.
The retirement notice
From Moonshot's official pricing page, verbatim:
"Dear Kimi Users, starting October 31, 2026, K2 series models will no longer be available to newly registered users; existing users are not affected. We plan to retire all K2 series models by the end of December 2026, please prepare your migration in advance."
Source: Moonshot pricing, chat models
- Oct 31, 2026: new accounts can't use K2.
- End of Dec 2026 (planned): all K2 models retired.
Both models discount cached input, so run your hit rate through the prompt caching calculator.
Kimi K2 price vs K3.1 Pro (per 1M tokens)
| Model | Input (cache miss) | Output | Cached input |
|---|---|---|---|
| kimi-k2.6 | $0.95 | $4.00 | $0.16 |
| kimi-k2.7-code | $0.95 | $4.00 | $0.19 |
| kimi-k3.1-pro | $2.00 | $10.00 | $0.20 |
Source: Moonshot pricing, chat models, checked Oct 24, 2026. K3.1 Pro also bills cache writes separately: $2.00 (5-minute TTL) or $4.00 (1-hour TTL) per 1M tokens.
The jump:
- Input: $2.00 / $0.95 = ~2.1x
- Output: $10.00 / $4.00 = 2.5x
- Cached input: +$0.04 (from K2.6) or +$0.01 (from K2.7 Code)
Worked example: a typical agent, per month
This uses the TokenTab calculator defaults and math: 3,000 input and 800 output tokens per call, 5 tool calls per task, 100 tasks/day and 30 days. Cost per task = calls × (input × input price + output × output price) / 1,000,000. No caching discounts are applied.
| K2.6 / K2.7 Code | K3.1 Pro | |
|---|---|---|
| Per task | 5 × (3,000 × $0.95 + 800 × $4) / 1M = $0.03025 | 5 × (3,000 × $2 + 800 × $10) / 1M = $0.0700 |
| Per day (100 tasks) | $3.025 | $7.00 |
| Per month (30 days) | $90.75 | $210.00 |
The same workload costs about 2.3x more on K3.1 Pro, or +$119.25/month. The blend lands between 2.1x and 2.5x because output tokens make up more than half of the K2 bill here.
Your ratio depends on your input/output mix. Input-heavy agents (long contexts, short answers) land closer to 2.1x. Output-heavy agents (code generation) land closer to 2.5x. Try your own numbers in the calculator.
FAQ
What is the Kimi K2 price?
On Moonshot's official API, kimi-k2.6 and kimi-k2.7-code both cost $0.95 per 1M input tokens and $4.00 per 1M output tokens. Cached input is $0.16 (K2.6) or $0.19 (K2.7 Code).
When is Kimi K2 being retired?
New users lose access from October 31, 2026, while existing users are not affected. Moonshot plans to retire all K2 series models by the end of December 2026.
How much more expensive is Kimi K3.1 Pro than K2?
About 2.1x on input ($2.00 vs $0.95) and 2.5x on output ($10.00 vs $4.00). For a typical agent workload, that comes to roughly 2.3x overall.
Is there a faster K3.1 Pro option?
Yes, kimi-k3.1-pro-highspeed costs $4.00 input / $20.00 output, with $0.20 cached input.
Can I still sign up and use K2 after October 31?
Not as a newly registered user, according to Moonshot's notice. Existing users keep access until the planned retirement.
Prices are from Moonshot's official pricing page and were checked October 24, 2026. Prices change, so confirm them before budgeting.