The linked official Moonshot AI lifecycle catalog lists the exact model ID kimi-k2.5 as retired on 2026-08-31; the official catalog confirms requests now return HTTP 404.
Historical
Common family use cases
Assumption · Read long documents · Work across languages · Read, analyze, or write
Unknown · No comparable public ranking signal is currently linked to this exact version.
Verify the exact version before choosing: capabilities, price, context limits, latency, tools, and regional availability can differ by provider access path.
Assumption · OpenGPT catalog guidance
Best for
Compatibility checks, controlled migrations, or reproducing an existing deployment.
Long-context, Chinese, reasoning, coding, and hosted agent workflows.
Strengths
A previous-generation role can support continuity while a migration is evaluated.
Kimi K2.5 is provider-listed as the previous-generation release in its model family.
Weaknesses
This release is retired as of 2026-08-31, so adoption requires a migration plan.
Verify exact capabilities, pricing, latency, context limits, tools, and regional availability with the provider.
Assumption · These points are OpenGPT decision guidance derived from provider-confirmed family, catalog role, and lifecycle information. They are not measured results for this exact version. Verify capabilities, pricing, latency, context limits, tools, and regional availability with the provider.
Model family
Family-level guidance
The following summary describes the broader model family from official positioning. It is not a measured result for this exact version.
KK
Assumption · Model family: Kimi
Kimi
Moonshot AI's hosted family positioned for long-context, Chinese, reasoning, and agent workflows.
Read long documentsWork across languagesRead, analyze, or writeUse current information or tools
Hosted APIs emphasize long documents and Chinese-language use. Provides reasoning and tool-oriented model options.
!
Family limitations
Closed hosted access limits deployment control. Availability, naming, and features can vary by region and platform version.
Verify the exact version before choosing: capabilities, price, context limits, latency, tools, and regional availability can differ by provider access path.
Provider documentation and OpenGPT catalog classification are labeled separately.
Version differences
Compare two catalog model IDs from one model family. Provider IDs, sources and published ranking signals are shown; OpenGPT catalog classification is labeled separately.
No verified feature-difference data is available. OpenGPT does not infer speed, price, context length, or quality.
OFFICIAL MODEL PROFILE
Official resources
These links go to the provider. Availability and the selectable version can vary by product, account, region, and endpoint.
OFFICIAL MODEL PROFILE
Published ranking signals
Each row preserves one published metric and source. Scores from different sources are not interchangeable.
0
No comparable public ranking signal is currently linked to this exact version.
Missing data is not a zero score and does not mean the model is weak. It means this edition cannot safely attach a published result to the exact official version.
OpenGPTAI decision workspace