Home / Providers / DeepInfra

DeepInfra

DeepInfra serves 22 of the 50 models Gimbix tracks most closely, 0 of them with token log probabilities.

ModelInput / output per 1MLog probabilitiesQuantization
Kimi K3$2.85 / $14.25nobf16
GLM 5.3$0.563 / $2.50nofp4
DeepSeek V4 Pro 0423$1.30 / $2.60nofp8
gpt-oss-120b$0.037 / $0.17nobf16
gpt-oss-120b$0.15 / $0.60nobf16
gpt-oss-120b$0.20 / $0.95nofp8
gpt-oss-20b$0.03 / $0.14nobf16
Gemma 4 31B$0.09 / $0.34nofp4
Gemma 4 31B$0.13 / $0.38nofp8
Gemma 4 31B$0.27 / $0.76nofp8
DeepSeek V4.1 Flash$0.14 / $0.42nofp8
DeepSeek V4 Flash 0731$0.06 / $0.18nofp8
GLM 5.3 Flash$0.075 / $0.25nofp4
MiMo-V2.6-Flash$0.14 / $0.28nofp8
Nemotron 3.5 Lightning$0.08 / $0.20nobf16
Nemotron 3 Ultra$0.50 / $2.20nofp4
Qwen3.8 27B$0.15 / $1.88nobf16
MiniMax M3$0.28 / $1.10nofp8
Kimi K2.6$0.75 / $3.50nofp4
Llama 4 Maverick$0.20 / $0.80nofp8
Llama 3.3 70B Instruct$0.10 / $0.32nofp8
Phi 4$0.07 / $0.14nobf16