text model comparison
Kimi K2.6 vs Kimi K3
Kimi K2.6 costs less than Kimi K3 at every mix of input and output — $0.55/M against $1.85/M on input and $2.30/M against $9.00/M on output — so at 100M input and 20M output tokens a month it comes to $101.00 against $365.00, a difference of $264.00, and what Kimi K3 holds over it is context rather than price — 512K against 256K, 2.0× wider.
Side by side
| Kimi K2.6 | Kimi K3 | |
|---|---|---|
| Our rate — input | $0.55/M | $1.85/M |
| Our rate — output | $2.30/M | $9.00/M |
| Our rate — cached input | $0.11/M | $0.37/M |
| Official rate — input | $0.95/M | $3.00/M |
| Official rate — output | $4.00/M | $15.00/M |
| Below official | 42% | 40% |
| Context window | 256K | 512K |
| Publisher | Moonshot AI | Moonshot AI |
| Licence | Modified MIT | Modified MIT |
| Released | Jul 18, 2026 | Aug 26, 2026 |
| Model id | moonshotai/kimi-k2.6 | moonshotai/kimi-k3 |
| Callable today | Yes | Yes |
What it costs, at 100M input and 20M output tokens a month
One volume, both models, this catalogue’s published rates. Your figure will differ; the arithmetic will not, and the ratio of the two columns is the part that transfers to your own numbers.
At this volume Kimi K2.6 is $264.00 cheaper than Kimi K3 — 72% less. That ordering holds at any ratio of input to output, because the cheaper model is cheaper on every rate rather than on one column.
Excludes cached-input savings, which on a repeated-prefix workload reduce the input side substantially. Cached rates are in the table above and on the pricing page.
When to pick each
Pick Kimi K2.6
- Input costs $0.55/M against Kimi K3's $1.85/M, which decides the bill on input-heavy work — long documents in, short answers out.
- Output costs $2.30/M against Kimi K3's $9.00/M, which is the whole bill on generation-heavy work where a short brief produces long copy.
- Cached input is $0.11/M against Kimi K3's $0.37/M, and on an agent loop the repeated prefix is most of every request.
- The catalogue lists it for long-horizon coding tasks and multi-file refactors, agent workflows that chain dozens of tool calls and repo-scale code review and migration passes.
Pick Kimi K3
- It holds 512K in one request against 256K — 2.0× wider, which decides whether a long input is one call or a map-reduce.
- The catalogue lists it for whole-codebase reasoning without a retrieval layer, research synthesis over hundreds of documents and complex multi-step planning agents.
Every line above is a field in the model catalogue rather than a judgement about output quality. This site publishes no benchmark results, so which model answers your prompts better is a measurement to run on your own traffic — the same key calls both.
Questions
- Which is cheaper, Kimi K2.6 or Kimi K3?
- Kimi K2.6, at 100M input and 20M output tokens a month: $101.00 against $365.00, a difference of $264.00 — 72% less. That holds at any ratio of input to output on this pairing, because Kimi K2.6 has the lower rate on every column, not just the one the headline quotes. What the arithmetic does not tell you is whether both models do your job to the same standard; that is a measurement on your own prompts, not a number we can publish.
- Can I call both Kimi K2.6 and Kimi K3 today?
- Yes — both. Requests naming `moonshotai/kimi-k2.6` or `moonshotai/kimi-k3` reach a live upstream today and are billed at the rates above. 7 of the 24 models in this catalogue are callable right now; the rest are listed at their intended price and marked, so you find out here rather than from an error at call time.
- How much context does Kimi K2.6 hold compared with Kimi K3?
- Kimi K3 takes 524,288 tokens (512K) in a single request against Kimi K2.6's 262,144 (256K) — 2.0× wider. The window is a cost decision before it is a capability one: an input that does not fit has to be split, summarised in pieces and summarised again, which bills the source text more than once. Below that line the wider window is capacity you are not using.
- What licences are Kimi K2.6 and Kimi K3 under?
- Both are Modified MIT — Kimi K2.6 from Moonshot AI and Kimi K3 from Moonshot AI. Licence is therefore not a differentiator on this pairing, which is worth knowing before it becomes the question a legal review opens with. Read the text itself before shipping either; a licence name is a pointer to terms, not a summary of them.
- How do I switch between Kimi K2.6 and Kimi K3?
- Change the model id and nothing else. Both are served from `https://router.xark.io/api/v1` behind the same OpenAI-compatible contract — `moonshotai/kimi-k2.6` and `moonshotai/kimi-k3` — so the request shape, the streaming frames, the usage block and the error envelope are identical between them. Send an explicit id and no substitution happens underneath you, which is what makes running both and comparing the two on your own traffic a change of one string rather than a migration.
- How do these rates compare with the official ones?
- Both rows are published side by side rather than as a single discount claim. Kimi K2.6 is 42% below Moonshot AI's official rate and Kimi K3 is 40% below Moonshot AI's, which at 100M input and 20M output tokens a month is $101.00 against an official $175.00 for Kimi K2.6, and $365.00 against $600.00 for Kimi K3. The claim on this site is only that our rate sits below the model publisher's own — not that no one else is cheaper. Where a third-party provider undercuts us on either model, that provider's rate is printed on that model's own page.