GLM 5.2vsKimi K2.5
Kimi K2.5 is cheaper than GLM 5.2 at $0.6/M vs $1.4/M input tokens.
| Spec | GLM 5.2 | Kimi K2.5 |
|---|---|---|
| Input price | $1.4/M tokens | $0.6/M tokens |
| Output price | $4.4/M tokens | $3/M tokens |
| Context window | 1.0M tokens | 262K tokens |
| Weights | Open | Open |
| Free API (OpenRouter) | No | No |
| Released | Jun 2026 | Jan 2026 |
Our Verdict
GLM 5.2
Kimi K2.5Runner-up
Not enough votes to call it. On the specs, GLM 5.2 has the edge: bigger model tier, newer, bigger context window, major provider backing.
Slight edge
Reviewing agent-written code?See a Brief PR report
API pricing
Cost per 1M tokens
GLM 5.2
Input
$1.40
Output
$4.40
Kimi K2.5
Input
$0.60
2.3× cheaper
Output
$3.00
1.5× cheaper
Kimi K2.5 is cheaper on both: 2.3× input, 1.5× output.
Where to run it
30 hosts, cheapest first
GLM 5.224 hosts
Baidu Qianfanfp8$0.48 in$1.51 out1M99.9% upStreamLakefp8$0.48 in$1.51 out1M99.4% upDeepInfrafp4$0.49 in$1.56 out1M99.4% upAmbientfp8$0.60 in$2.00 out203k98.2% upNovitafp8$0.68 in$2.15 out1M99.9% upDigitalOcean$0.70 in$2.20 out262k99.3% up
18 more hostsFewer hosts
CoreWeavefp4$0.76 in$2.42 out1M99.8% upAtlasCloudfp8$0.94 in$2.95 out1M99.9% upAlibaba Cloudfp8$0.97 in$3.04 out1M99.9% upInceptronfp4$1.08 in$2.89 out1M99.2% upSiliconFlowfp8$1.19 in$3.74 out1M100% upPhalafp8$1.26 in$3.00 out1M99.6% upBasetenfp8$1.40 in$4.40 out1M100% upCloudflare Workers AI$1.40 in$4.40 out262k100% upCrusoefp8$1.40 in$4.40 out1M98.1% upFireworks$1.40 in$4.40 out1M100% upFriendli$1.40 in$4.40 out1M99.9% upGMI Cloudfp8$1.40 in$4.40 out1M99.8% upMistral$1.40 in$4.40 out1M100% upParasailfp4$1.40 in$4.40 out262k99.9% upTogether$1.40 in$4.40 out512k98.6% upVenicefp8$1.40 in$4.40 out1M99.6% upZ.aifp8$1.40 in$4.40 out1M99.9% upDecartfp4$2.10 in$6.60 out1M99.8% up
Per million tokens. Prices and uptime via OpenRouter, checked 13 Sep 2026.
Writing DNA
Style Comparison
Similarity
47%
GLM 5.2
Kimi K2.5
56%Vocabulary62%
18wSentence Length22w
0.27Hedging0.50
3.7Bold4.2
2.8Lists2.4
0.00Emoji0.01
0.56Headings0.51
0.24Transitions0.17
Based on 27 + 26 text responses
Research
What we learned reading every model
FAQ
Common questions
Keep exploring
