All times Eastern (ET)Provider pages checked daily at

Kimi K2.5

API ID kimi-k2.5

kimi-k2.5 was a Moonshot AI model, released January 2026.

Status
Retired
Released
January 2026
Shutdown
Aug 31, 2026
Replacement
Kimi K3kimi-k3

kimi-k2.5 stopped serving on August 31, 2026. Recommended replacement: kimi-k3.

How to migrate to Kimi K3

Benchmarks

Scores as each source published them
or

Comparing with Kimi K2 0711 Preview, the previous model in this line. A score is compared only with the same benchmark from the same source, at the same setting where both have it. Full comparison: prices, limits, features

Arena

ArenaKimi K2.5Kimi K2 0711 PreviewChangeRankDateSource
Text
Other settings
  • 1431 ±7 vs not rated no thinking
By category
  • Math 1471 rank 53 vs 1388 (+83) at thinking (vs no setting)
  • Creative writing 1423 rank 80 vs 1371 (+52) at thinking (vs no setting)
  • Instruction following 1439 rank 82 vs 1381 (+58) at thinking (vs no setting)
  • Hard prompts 1471 rank 83 vs 1432 (+39) at thinking (vs no setting)
  • Long prompts 1458 rank 78 vs 1395 (+63) at thinking (vs no setting)
  • Multi-turn 1452 rank 85 vs 1419 (+33) at thinking (vs no setting)
  • Expert 1482 rank 69 vs 1415 (+67) at thinking (vs no setting)
1450 ±31418 ±5+3279 of 415
74,777 votes
Oct 13, 2026source
Text: Coding
Other settings
  • 1502 ±5 vs not rated thinking
1504 ±141460 ±8+4469 of 410
1,874 votes
Oct 13, 2026source
WebDev
Other settings
  • 1405 ±12 vs not rated no thinking
1436 ±5not rated78 of 142
20,135 votes
Oct 13, 2026source
Vision
Other settings
  • 1236 ±11 vs not rated no thinking
1252 ±6not rated51 of 165
28,297 votes
Oct 9, 2026source
Document1436 ±7not rated37 of 44
21,569 votes
Sep 13, 2026source

Rated as kimi-k2.5-thinking; Kimi K2 0711 Preview as kimi-k2-0711-preview. Some were rated at different settings; ratings on one arena share one scale, so the difference still reads directly. Ratings come from people voting between two anonymous answers; Text, Vision and Document are style-controlled, as on arena.ai. Data: Arena (arena.ai), CC BY 4.0.

Epoch AI

BenchmarkKimi K2.5Kimi K2 0711 PreviewChangeSettingRunSource
SimpleQA Verified34.3%not runAug 27, 2026source
SWE-bench Verified73.8%not runFeb 17, 2026source

Run independently by Epoch AI. Data: Epoch AI, Capabilities & benchmarking (epoch.ai/benchmarks), CC BY.

On cloud platforms

All platform dates
PlatformPlatform IDStateShutdownTime leftReplacementSource
Amazon Bedrockmoonshotai.kimi-k2.5Activenot before Jan 27, 2027—source
Microsoft Foundryversion 1Kimi-K2.5ActiveJan 26, 2027103 days—source

Microsoft Foundry (version 1) retires this model 148 days after Moonshot AI's own API.

Read daily from each platform's own pages. The shutdown date and countdown on this page use Moonshot AI's own date. Time left counts to 00:00 UTC on the shutdown date, which is the evening before in ET.

Price history

USD per 1M tokens, standard tier

No price history yet.

Prices

USD per 1M tokens, every type and band, by tier

No published prices.

Changelog

Changes we detected on Moonshot AI's pages

No changes recorded yet.

Last checked · Sources: Moonshot AI pricing, Moonshot AI batch, Moonshot AI models