GLM 5.2vsQwen3.5 122B A10B
Qwen3.5 122B A10B is cheaper than GLM 5.2 at $0.4/M vs $1.4/M input tokens.
| Spec | GLM 5.2 | Qwen3.5 122B A10B |
|---|---|---|
| Input price | $1.4/M tokens | $0.4/M tokens |
| Output price | $4.4/M tokens | $3.2/M tokens |
| Context window | 1.0M tokens | 262K tokens |
| Weights | Open | Open |
| Free API (OpenRouter) | No | No |
| Released | Jun 2026 | Feb 2026 |
Our Verdict
GLM 5.2
Qwen3.5 122B A10BRunner-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
API pricing
Cost per 1M tokens
GLM 5.2
Input
$1.40
Output
$4.40
Qwen3.5 122B A10B
Input
$0.40
3.5× cheaper
Output
$3.20
1.4× cheaper
Qwen3.5 122B A10B is cheaper on both: 3.5× input, 1.4× output.
Where to run it
29 hosts, cheapest first
GLM 5.224 hosts
StreamLakefp8$0.28 in$0.88 out1M98.9% upBaidu Qianfanfp8$0.49 in$1.53 out1M99.9% upDeepInfrafp4$0.49 in$1.56 out1M98.9% upAmbientfp8$0.60 in$2.00 out203k97.9% upNovitafp8$0.68 in$2.15 out1M99.9% upDigitalOcean$0.70 in$2.20 out262k99.6% 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 out1M99.5% upSiliconFlowfp8$1.19 in$3.74 out1M99.9% upInceptronfp4$1.25 in$2.99 out1M99.1% upPhalafp8$1.26 in$3.00 out1M99.6% upBasetenfp8$1.40 in$4.40 out1M99% upCloudflare Workers AI$1.40 in$4.40 out262k100% upCrusoefp8$1.40 in$4.40 out1M99.9% upFireworks$1.40 in$4.40 out1M99.9% upFriendli$1.40 in$4.40 out1M99.8% upGMI Cloudfp8$1.40 in$4.40 out1M99.4% upMistral$1.40 in$4.40 out1M99.6% upParasailfp4$1.40 in$4.40 out262k99.8% upVenicefp8$1.40 in$4.40 out1M99.7% upZ.aifp8$1.40 in$4.40 out1M99.8% upDecartfp4$2.10 in$6.60 out1M99.7% upTogetherdegraded$1.40 in$4.40 out512k97.8% up
Per million tokens. Prices and uptime via OpenRouter, checked 9 Sep 2026.
Writing DNA
Style Comparison
Similarity
68%
Qwen3.5 122B A10B uses 10.4x more emoji
GLM 5.2
Qwen3.5 122B A10B
56%Vocabulary49%
18wSentence Length15w
0.27Hedging0.29
3.7Bold6.2
2.8Lists4.8
0.00Emoji0.10
0.56Headings0.88
0.24Transitions0.20
Based on 27 + 24 text responses
Research
What we learned reading every model
FAQ
