Skip to content

Most Reliable Qwen2.5 72B Instruct Inference Providers

Ranks providers by the uptime OpenRouter measured during the latest 30-minute window.

DeepInfra currently ranks first at 99.75% via deepinfra/fp8.

Provider options:
2
Ranked:
2
Metric:
Uptime (last 30 min)

Highest uptime endpoint ranking

Most Reliable Qwen2.5 72B Instruct Inference Providers
RankProvider / routeUptime (last 30 min)Speed measurementsEndpoint details
#1Provider / route
DeepInfradeepinfra/fp8
Uptime (last 30 min)99.75%Speed measurements
500-token response
25.49 s
Response starts
0.49 s
Output speed
20.0 tok/s
Endpoint details
Input price
$0.36 / 1M tokens
Output price
$0.4 / 1M tokens
Uptime (30 min)
99.75%
Context
32.8K
#2Provider / route
NovitaAInovita/bf16
Uptime (last 30 min)72.08%Speed measurements
500-token response
29.10 s
Response starts
7.36 s
Output speed
23.0 tok/s
Endpoint details
Input price
$0.38 / 1M tokens
Output price
$0.4 / 1M tokens
Uptime (30 min)
72.08%
Context
32K

Endpoint data fetched .

Qwen2.5 72B Instruct endpoint guide

How to interpret the Most Reliable Qwen2.5 72B Instruct Inference Providers

2 of 2 Qwen2.5 72B Instruct endpoints currently have the published data required for this ranking. DeepInfra leads at 99.75% via deepinfra/fp8. The table keeps unranked routes visible so missing measurements do not look like missing provider availability.

How recent uptime is ranked

The reliability ranking uses the published 30-minute endpoint uptime measurement. This short operational window can reveal current instability, but it is not a service-level agreement or a substitute for longer-term availability monitoring.

Compare DeepInfra with the next option

NovitaAI currently ranks second at 72.08%. Compare that gap with input and output price, response-start time, output speed, 30-minute uptime, context length, and the exact route before deciding whether first place is meaningful for your workload.

Use the route, not only the provider name

Qwen2.5 72B Instruct has 2 published provider options, and performance data is matched to each exact OpenRouter routing tag. Different quantization, context, regional deployment, or provider configuration can change price and behavior even when the underlying model name is identical.