Model comparisons
“Which of these two is cheaper” almost never has a single answer: one model wins on input, the other on output, and which one you should run depends on the shape of your traffic. Each page below prices both models at one stated volume and says where the answer flips.
23 comparisons, all within a modality — comparing a video model against an embedding model is not a comparison, and their prices are not even in the same unit. Every pairing includes at least one model that can be called today, and says plainly when the other cannot.
Text models
Kimi K2.6 vs Kimi K3
$0.55/M against $1.85/M on input. At 100M input and 20M output tokens a month, Kimi K2.6 costs $101.00 against $365.00.
Kimi K2.6 vs GLM-5.2
$0.55/M against $0.82/M on input. At 100M input and 20M output tokens a month, Kimi K2.6 costs $101.00 against $133.00.
Kimi K2.6 vs DeepSeek V4 Pro
$0.55/M against $0.28/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Pro costs $39.00 against $101.00.
Kimi K2.6 vs DeepSeek V4 Flash
$0.55/M against $0.090/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Flash costs $12.60 against $101.00.
Kimi K3 vs GLM-5.2
$1.85/M against $0.82/M on input. At 100M input and 20M output tokens a month, GLM-5.2 costs $133.00 against $365.00.
Kimi K3 vs DeepSeek V4 Pro
$1.85/M against $0.28/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Pro costs $39.00 against $365.00.
Kimi K3 vs DeepSeek V4 Flash
$1.85/M against $0.090/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Flash costs $12.60 against $365.00.
GLM-5.2 vs DeepSeek V4 Pro
$0.82/M against $0.28/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Pro costs $39.00 against $133.00.
GLM-5.2 vs DeepSeek V4 Flash
$0.82/M against $0.090/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Flash costs $12.60 against $133.00.
DeepSeek V4 Pro vs DeepSeek V4 Flash
$0.28/M against $0.090/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Flash costs $12.60 against $39.00.
Kimi K2.6 vs GLM-5.2 Air
One servable$0.55/M against $0.17/M on input. At 100M input and 20M output tokens a month, GLM-5.2 Air costs $29.40 against $101.00.
Kimi K2.6 vs Qwen3 Max Instruct
One servable$0.55/M against $0.49/M on input. At 100M input and 20M output tokens a month, Qwen3 Max Instruct costs $88.00 against $101.00.
Kimi K3 vs GLM-5.2 Air
One servable$1.85/M against $0.17/M on input. At 100M input and 20M output tokens a month, GLM-5.2 Air costs $29.40 against $365.00.
Kimi K3 vs Qwen3 Max Instruct
One servable$1.85/M against $0.49/M on input. At 100M input and 20M output tokens a month, Qwen3 Max Instruct costs $88.00 against $365.00.
GLM-5.2 vs GLM-5.2 Air
One servable$0.82/M against $0.17/M on input. At 100M input and 20M output tokens a month, GLM-5.2 Air costs $29.40 against $133.00.
GLM-5.2 vs Qwen3 Max Instruct
One servable$0.82/M against $0.49/M on input. At 100M input and 20M output tokens a month, Qwen3 Max Instruct costs $88.00 against $133.00.
GLM-5.2 Air vs DeepSeek V4 Pro
One servable$0.17/M against $0.28/M on input. At 100M input and 20M output tokens a month, GLM-5.2 Air costs $29.40 against $39.00.
DeepSeek V4 Pro vs Qwen3 Max Instruct
One servable$0.28/M against $0.49/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Pro costs $39.00 against $88.00.
GLM-5.2 Air vs DeepSeek V4 Flash
One servable$0.17/M against $0.090/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Flash costs $12.60 against $29.40.
DeepSeek V4 Flash vs Qwen3 Max Instruct
One servable$0.090/M against $0.49/M on input. At 100M input and 20M output tokens a month, DeepSeek V4 Flash costs $12.60 against $88.00.
Image models
FLUX.2 [schnell] vs Qwen-Image
One servable$0.0018/img against $0.0120/img on input. At 10,000 generated images, FLUX.2 [schnell] costs $18.00 against $120.00.
Qwen-Image vs HiDream-I1
One servable$0.0120/img against $0.0150/img on input. At 10,000 generated images, Qwen-Image costs $120.00 against $150.00.
Embedding models
No comparison here ranks one model above another on quality. This site publishes no benchmark results, so the only claims it makes about two models are the ones the catalogue can evidence: the rates, the context window, the licence, and whether a request would reach an upstream today. Per-model detail is on the model pages.