Kimi K2.5 · Kimi K2.6 · Kimi K3 · Kimi K3 Fast
Moonshot model family
Use Moonshot Kimi across two generations, with the differences kept explicit.
Vizify lists four Kimi chat variants from Moonshot: K3 and K3 Fast from the newer K3 generation, and K2.5 and K2.6 from the K2 generation. K3 succeeds the K2 models. All four require Pro or a credit pack purchase and accept image input, and reasoning control differs by generation.
Four Kimi variants across two generations: K3, K3 Fast, K2.6, and K2.5 Kimi K3 succeeds K2.5 and K2.6, and replaces the retired K2.7 Code for coding work All four Kimi models require paid access All four accept text and image input K2-generation reasoning is a single on/off toggle; K3-generation reasoning is fixed with no selectable preset One selected Kimi model per request
Verified catalog facts
What Vizify exposes for this model family.
These facts are generated from the current public catalog, not copied from provider marketing.
Text · Image
Text
Chat
Reviewed 2026-09-26
Where this family fits.
Opening Kimi K3, the newest generation, directly from this page
Coding and long-horizon work that previously ran on Kimi K2.7 Code
Moving a K2.5 or K2.6 workflow to K3 for your next request
Turning reasoning on or off on K2.5 or K2.6 when a simple toggle is enough
Requests that combine text with an uploaded image on any Kimi variant
What not to assume.
K3, K3 Fast, K2.6, and K2.5 all require paid access
Kimi K2.7 Code is no longer offered in Vizify
K3 and K3 Fast use a fixed reasoning behavior with no selectable on/off or effort preset
K2.5 and K2.6 expose only an on/off reasoning toggle, not a graded effort scale
Vizify sends each request to one selected Kimi variant rather than running several at once
Two generations under one family
Vizify currently exposes four Kimi chat variants from Moonshot, spanning two generations. The newer K3 generation includes K3 and K3 Fast. The K2 generation includes K2.5 and K2.6. All four variants are in the Paid catalog.
Kimi K3 is the successor
Kimi K3 succeeds K2.5 and K2.6. Moonshot positions it as its flagship for long-horizon coding and end-to-end knowledge work. The K2 models stay available, so existing conversations keep working; pick K3 or K3 Fast for your next request to move to the newer generation.
Kimi K2.7 Code was deprecated upstream in September 2026 and has been removed from Vizify. If you used it for coding, choose Kimi K3 instead. K3 also accepts image input, which K2.7 Code did not.
Image input on every variant
Kimi K3, K3 Fast, K2.5, and K2.6 are all in the Vizify vision list, so the customer App can pass a supported image upload along with the request after it validates file and account requirements.
Reasoning control changes with the generation
The two Kimi generations handle reasoning differently. K2.5 and K2.6 expose a single on/off toggle: reasoning is either applied or it is not, with no graded effort scale in between. K3 and K3 Fast use a fixed, model-managed reasoning behavior with no selectable preset at all. Check the model picker in Vizify for the control that applies to the variant you choose.
Capability review
Availability, tier, image-input support, and reasoning control were checked against the Vizify repository and app profile on 2026-09-26, and Kimi K2.7 Code’s upstream deprecation against the Vercel AI Gateway notice. Search demand for these specific models has not been validated: the most recent keyword research package predates them, so this page is published on capability evidence alone and does not imply demand-backed priority. Provider benchmarks, external release status, pricing, and performance rankings are outside this page’s factual contract.
FAQ
Questions about Moonshot Kimi models in Vizify.
Which Moonshot Kimi variants are available in Vizify?
Four Kimi chat variants from Moonshot: K3 and K3 Fast from the K3 generation, and K2.5 and K2.6 from the K2 generation. All four are listed in the Paid tier.
What happened to Kimi K2.7 Code?
Kimi K2.7 Code was deprecated upstream in September 2026 and is no longer offered in Vizify. For coding and long-horizon work, choose Kimi K3, which Moonshot positions as its flagship for that work, or K3 Fast. Unlike K2.7 Code, both also accept image input.
Which Kimi model does this page open?
The primary action opens Vizify with Kimi K3 as a one-time launch selection. The App validates that model against its live catalog before use.
Is Kimi K3 the default model in Vizify?
No. The account-wide default chat model is GPT-5.4 Mini. Selecting Kimi K3 from this page is an explicit one-time launch hint, not a change to your account default.
Are Kimi K2.5 and K2.6 still available?
Yes. K3 is their successor, but K2.5 and K2.6 stay in the Vizify catalog, so existing conversations keep working. Choose K3 for your next request when you want the newer generation.
How does reasoning control differ between Kimi generations?
Kimi K2.5 and K2.6 expose a single on/off reasoning toggle. Kimi K3 and K3 Fast use a fixed, model-managed reasoning behavior with no selectable on/off or effort preset.
Do all Kimi variants require Paid-model access?
Yes. Every Kimi variant requires Pro or a credit pack purchase. The App checks access before each request.
Does Vizify run multiple Kimi variants on the same request?
No. Vizify selects one model for each request. You can switch Kimi variants between requests, but this page does not promise parallel answers or automatic comparison.