Kimi K2.6 API Pricing
Moonshot (Kimi) · context window 262K · prices updated 2026-09-15
Where Kimi K2.6 sits on price
At $4.00 per million output tokens, Kimi K2.6 is a mid tier model — 1.9× the median output price of $2.08, which puts it cheaper than 37% and pricier than 63% of the models we track. It sits 6th cheapest of the 8 Moonshot (Kimi) models we track. Output costs 4.2× input, the usual spread — trim both, starting with the answer length.
Cached input is 83% cheaper. On an input-heavy workload you need roughly a 30% cache-hit rate to take 25% off the input line — reachable for chatbots and agents that resend the same system prompt and history.
What real workloads cost on Kimi K2.6
These three workloads are the ones teams actually run on a mid-priced model — a $4.00/1M model is not bought for the same job as one ten times the price.
| Workload | Per request | Per month |
|---|---|---|
| Support chatbot — 500 conversations/day (5K in / 1.4K out) Conversation history re-sent each turn — the classic prompt-caching win. | $0.0103 | $155 |
| Document pipeline — 10K docs/day (3K in / 500 out) Batch-friendly: no user waiting, so a ~50% batch discount usually applies. | $0.0049 | $1455 |
| Small app — 1K requests/day (500 in / 150 out) The baseline most side projects actually run at. | $0.0011 | $32.25 |
Model your exact traffic in the API cost calculator — it preloads Kimi K2.6 with caching and batch options.
Kimi K2.6 price history
Moonshot (Kimi) last raised Kimi K2.6's output price on Aug 23, 2026 (+75% per output token), from $0.5415/$2.28 to $0.95/$4.00 per 1M in/out. That is one of 12 repricings in the 44 days we have tracked it — worth knowing before you hard-code today's rate into a budget.
| Date | Input /1M | Output /1M | Output change |
|---|---|---|---|
| Aug 2, 2026 (tracking began) | $0.6 | $3.41 | — |
| Aug 4, 2026 | $0.589 | $2.48 | -27% |
| Aug 6, 2026 | $0.57 | $2.40 | -3% |
| Aug 7, 2026 | $0.589 | $2.48 | +3% |
| Aug 8, 2026 | $0.5795 | $2.44 | -2% |
| Aug 10, 2026 | $0.95 | $4.00 | +64% |
| Aug 14, 2026 | $0.5605 | $2.36 | -41% |
| Aug 15, 2026 | $0.65 | $3.41 | +44% |
| Aug 16, 2026 | $0.5415 | $2.28 | -33% |
| Aug 17, 2026 | $0.95 | $4.00 | +75% |
| Aug 21, 2026 | $0.5795 | $2.44 | -39% |
| Aug 22, 2026 | $0.5415 | $2.28 | -7% |
| Aug 23, 2026 | $0.95 | $4.00 | +75% |
Change-points from our daily price snapshots (tracking since Aug 2, 2026; intraday moves between snapshots are not captured).
Kimi K2.6 vs Qwen3 VL 235B A22B Thinking
The closest-priced alternative from another vendor is Qwen3 VL 235B A22B Thinking (Alibaba (Qwen)) — priced within a rounding error on output, with $0.55 less per million input tokens. When two models land this close on price, the decision is quality on your own workload, not the price sheet: run 50 real requests through both and compare.
See Qwen3 VL 235B A22B Thinking pricing →Cheaper alternatives
More Moonshot (Kimi) models
Frequently asked questions
› How much does the Kimi K2.6 API cost?
Kimi K2.6 costs $0.95 per million input tokens and $4.00 per million output tokens, with cached input at $0.16 per million (83% cheaper). That works out to roughly 250,000 output tokens per dollar.
› What does the support chatbot workload cost on Kimi K2.6?
Support chatbot — 500 conversations/day (5K in / 1.4K out) costs about $0.0103 per request and $155 per month on Kimi K2.6. Conversation history re-sent each turn — the classic prompt-caching win.
› What does it cost to fill Kimi K2.6's 262K context window?
Sending 262K of input in a single request costs $0.249 at $0.95 per million tokens — before any output. With prompt caching that same fill drops to about $0.0419 on repeat requests. This is why large context windows are cheap to advertise and expensive to actually use.
› Is Kimi K2.6 worth the price?
Kimi K2.6 sits in the middle of the market (6 of 8 by price within Moonshot (Kimi)). The honest test is a routing experiment: send the same 200 real requests to this model and to a tier below, and compare failure rate against the price gap — most teams find a majority of traffic never needed the pricier model.