Qwen3.8 27B API pricing on Deep Infra
Every Deep Infra Qwen3.8 27B rate we track, compared against 21 providers serving the same model.
The cheapest Qwen3.8 27B input price we currently track is $0.080/1M on VoltageGPU.
The cheapest Qwen3.8 27B output price is $0.500/1M on Empirio Labs.
Deep Infra Qwen3.8 27B rates
Prices per 1M tokens. Batch and cached-input rates are shown where the provider publishes them.
| Mode | Input | Output |
|---|---|---|
| Standard | $0.150/1M | $1.88/1M |
About Qwen3.8 27B
- Creator
- Alibaba
- Modalities
- text
- Open weights
- Yes
Qwen3.8 27B on other providers
- VoltageGPU$0.080/1M
- ParalonCloud$0.140/1M
- OpenRouter$0.150/1M
- Runcrate$0.158/1M
- Empirio Labs$0.170/1M
- Geodd$0.200/1M
- AkashML$0.225/1M
- Crusoe$0.250/1M
- EcoHash$0.300/1M
- IO.NET$0.388/1M
- Lyceum$0.400/1M
- Novita AI$0.420/1M
- GMI Cloud$0.450/1M
- Prime Intellect$0.450/1M
- RouterPlus$0.450/1M
- OVHcloud$0.470/1M
- Scaleway$0.663/1M
- Groq$0.800/1M
- Cerebras$0.990/1M
- Heabsy$1.05/1M
About Deep Infra
Deep Infra is an AI inference cloud that provides serverless APIs for 100+ models and dedicated GPU rentals. Features OpenAI-compatible endpoints, custom model deployments, and infrastructure optimized for scale.
Frequently Asked Questions
How much does Qwen3.8 27B cost on Deep Infra?
Deep Infra serves Qwen3.8 27B from $0.150 per 1M input tokens, across 1 tracked pricing mode. Prices are collected daily — see the table above for current input, output, and cached-input rates.
Is Deep Infra the cheapest way to run Qwen3.8 27B?
Deep Infra ranks #3 of 21 providers we track serving Qwen3.8 27B on input price. The cheapest input price right now is on VoltageGPU. The lowest output price is $0.500/1M on Empirio Labs. Rates change often, and throughput, latency and context limits differ between providers — check the comparison above before committing.
Does Deep Infra offer batch or cached pricing for Qwen3.8 27B?
We track 1 Qwen3.8 27B offering from Deep Infra. Batch and cached-input rates appear in the table above when the provider publishes them; blank cells mean we have no current figure for that mode.