Kimi AI is free to start, with four paid plans for heavier use. This guide explains exactly what each plan includes, how Kimi's credits work, how much yearly billing saves, and what the Kimi API costs for developers, so you only pay for what you need.
Yes. Kimi AI has a free plan, and you do not need a credit card.
The free plan, Adagio, includes unlimited chat with Kimi K2.6 and a small monthly allowance for agent tasks and tools. Paid plans start at $19 a month ($15 a month billed yearly) and add more credits, faster priority, Agent Swarm, Kimi Code and heavier use of the newest model, Kimi K3.1 Pro.
Unlimited K2.6 chat plus about 6 agent credits a month.
Moderato: about 60 agent credits and Kimi Code.
From $48 to $480 a year, depending on the plan.
Kimi K3.1 Pro on the API, billed separately.
Kimi's plans are named after musical tempos, from slow (Adagio) to very fast (Vivace). Each paid plan includes everything in the plan before it.
Prices in US dollars from Kimi's help centre. Agent credits are Kimi's estimate of typical tasks per month. Mainland China plans are priced separately in yuan.
Every feature side by side. Scroll sideways on a phone.
| Feature | Adagio | Moderato | Allegretto | Allegro | Vivace |
|---|---|---|---|---|---|
| Monthly price | $0 | $19 | $39 | $99 | $199 |
| Billed yearly | Free | $180 ($15/mo) | $372 ($31/mo) | $948 ($79/mo) | $1,908 ($159/mo) |
| Chat with Kimi K2.6 | Free, unmetered | Free, unmetered | Free, unmetered | Free, unmetered | Free, unmetered |
| Agent credits a month | ~6 | ~60 | ~150 | ~360 | ~720 |
| Agent tasks at once | 1 | 2 | 2 | 4 | 4 |
| Speed priority | — | 4× | 4× | 4× | 4× |
| Agent Swarm | — | 25 uses · 2 subtasks | 50 uses · 4 subtasks | 120 uses · 4 subtasks | 240 uses · 8 subtasks |
| Kimi Code | — | 1× credits | 5× + HighSpeed | 15× + HighSpeed | 30× + HighSpeed |
| Kimi Claw | — | — | ✓ | ✓ | ✓ |
| Professional data calls | 200 | 2,000 | 5,000 | 12,000 | 24,000 |
US dollar prices from Kimi's help centre. Agent credits are Kimi's estimate of typical tasks. Check kimi.com before you pay, as plans can change.
Features draw from one shared pool that refills each billing cycle. K2.6 chat is free, and Kimi Code has its own pool.
Answer three quick questions and we'll suggest a starting plan. You can always upgrade later.
Monthly agent credits grow with each plan, from about 6 on Adagio to about 720 on Vivace.
Paying for a year up front saves about 20% on every paid plan. It's worth it once you know you'll keep using Kimi.
| Plan | Monthly price | 12 months, paid monthly | Paid yearly | You save |
|---|---|---|---|---|
| Moderato | $19 | $228 | $180 | $48 |
| Allegretto | $39 | $468 | $372 | $96 |
| Allegro | $99 | $1,188 | $948 | $240 |
| Vivace | $199 | $2,388 | $1,908 | $480 |
Kimi Code has its own pool with 5-hour and weekly limits. It runs on K2.8 Preview by default; HighSpeed models are 2–3× faster and use about twice the quota. Kimi Code guide
Kimi Claw is an always-on cloud agent with long-term memory and scheduled automation, including Claw for Android. It's included from Allegretto ($39) upward and uses your shared credits.
Calls to professional databases, such as financial data providers, are counted separately: 200 a month free, rising to 24,000 on Vivace.
Developers building their own apps pay per token, separately from any membership. There's no monthly fee: top up a balance on platform.kimi.ai and use it as you go.
| Model | Input | Output | Cached input |
|---|---|---|---|
kimi-k3.1-pro | $2.00 | $10.00 | $0.20 |
| K3.1 Pro Highspeed | $4.00 | $20.00 | — |
kimi-k3 | $3.00 | $15.00 | $0.30 |
kimi-k2.7-code | $0.95 | $4.00 | $0.19 |
kimi-k2.6 | $0.95 | $4.00 | $0.16 |
US dollars per million tokens. Thinking counts as output. Web Search API: $0.002–$0.003 per call. Batch jobs cost 60% of the standard price on K2.7 Code and K2.6. Kimi API guide
Uses standard list prices without caching or batch discounts. Real bills depend on how much the model thinks, which counts as output.
Use Adagio until you actually hit its limits. K2.6 chat never runs out.
Once you know your plan, yearly billing saves about 20%.
Use quick chat for simple questions and save agents and Deep Research for big jobs.
Free in the app, and the cheapest model on the API alongside K2.7 Code.
On the API, repeated input is billed at the much lower cached rate.
API batch jobs cost 60% of standard price on K2.7 Code and K2.6.