# Azure

HQ US · serves 17 tracked open-weights models, 1 confirmed by its own catalog. status page · privacy policy

models
17
verified
1
of overall top 10
4
cheapest on
3 models
vs best $
3.20×
ctx (median)
144k
uptime
73.1%

# models

model rank $/M in $/M out $/M cache src
deepseek-v4-pro #2 $1.91 $3.83 $0.16 openrouter
kimi-k2.6 #4 $0.95 $4.00 modelsdev
kimi-k2.7-code #7 $0.95 $4.00 $0.19 modelsdev
deepseek-v4-flash #8 $0.21 $0.56 $0.03 openrouter
kimi-k2.5 #12 $0.60 $3.00 modelsdev
deepseek-v3.2 #19 $0.58 $1.68 modelsdev
deepseek-r1 #33 $1.35 $5.40 modelsdev
gpt-oss-120b #35 $0.15 $0.60 azure
deepseek-v3.1 #37 $1.23 $4.94 litellm
deepseek-v3 #42 $1.14 $4.56 litellm
deepseek-v4-flash-0731 $0.44 $1.32 litellm
llama-3.3-70b-instruct $0.71 $0.71 modelsdev
mistral-large-3 $0.50 $1.50 litellm
phi-3.5-mini-instruct $0.13 $0.52 litellm
phi-4 $0.12 $0.50 modelsdev
phi-4-multimodal-instruct $0.08 $0.32 litellm
phi-4-reasoning-plus $0.12 $0.50 modelsdev

# recent changes

when (UTC)modelchange
Sep 4 09:43 deepseek-v4-flash-0731 $/M input $0.19 → $0.44 ↑ 131.6%
Sep 4 09:43 deepseek-v4-flash-0731 $/M output $0.51 → $1.32 ↑ 158.8%
Sep 2 09:42 deepseek-v4-flash-0731 now listed (via litellm)

all providers · machine-readable: providers.json