News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise ExpansionDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise Expansion
EveryNeocloud
Avian

Avian

Serverless and dedicated GPU inference for open-source models

Avian provides fast inference APIs and reserved GPU nodes for open-source large language models. Based in New York, it runs on Nvidia Blackwell chips within Microsoft Azure infrastructure, offering pay-per-token pricing and bare-metal reservations.

Share on WhatsApp

  • Runs on Nvidia B200 and B300 Blackwell GPUs
  • Hosted on SOC/2 approved Microsoft Azure infrastructure
  • Offers prepaid dedicated 8-GPU HGX nodes
  • Pay-per-token pricing from $0.1275 per million
  • Zero data retention and GDPR compliant

What it is

Avian is a provider of artificial intelligence inference services based in New York. It offers two main ways to access computing power: a serverless application programming interface that charges per token, and reserved bare-metal graphics processing unit nodes. The company focuses on hosting frontier open-source large language models, making them accessible to developers and enterprises without the need for complex infrastructure management. Its platform is designed to be a direct replacement for existing tools, requiring minimal code changes to integrate.

How it came about

The company states it was among the first to deploy the DeepSeek R1 model at scale when it launched in January 2025. Avian Data Inc. is incorporated in Delaware and headquartered in New York. It positions itself as a solution to the high cost and difficulty of deploying advanced AI models, aiming to handle the underlying infrastructure so users can focus on building applications. The service has expanded to include various recent open-source models and dedicated hardware options.

The compute

All models run on the latest Nvidia B200 and B300 Blackwell graphics processing units. The infrastructure uses speculative decoding and custom optimizations to maximize throughput and reduce latency. Users can access these resources via an OpenAI-compatible endpoint, which supports features like vision analysis and tool calling. For heavier workloads, Avian offers dedicated eight-GPU HGX nodes with full NVLink domains, ensuring single-tenant performance and predictable speeds without cold starts.

Where the GPUs are

The computing infrastructure is hosted on Microsoft Azure, leveraging its global network of data centres. This setup allows Avian to deploy inference endpoints across multiple regions, ensuring low-latency responses for users in different locations. The company states that its infrastructure is SOC/2 approved and complies with GDPR and CCPA regulations, with a policy of zero data retention after processing.

Customers and deals

The platform targets developers and enterprises who need scalable AI inference. It integrates with coding tools such as Claude Code, Cursor, Cline, and Kilo Code. Enterprise clients can request custom rate limits, volume pricing, or private deployments. The service is designed for both prototyping side projects and running production workloads, offering a flexible billing model that scales with usage.

Money and ownership

Avian operates on a pay-per-token model for its API, with prices starting from $0.1275 per million tokens. There are no monthly subscriptions or minimum commitments for this tier. For dedicated hardware, customers prepay for terms ranging from 12 months to three years, locking in their rates from the start. This dual approach allows teams to begin with serverless access and transition to reserved metal as their needs grow.

Its place in AI

Avian plays a role in making open-source AI models more accessible by handling the complex deployment infrastructure. By focusing on inference speed and cost efficiency, it supports the broader ecosystem of developers building with models like DeepSeek, MiniMax, and GLM. Its early adoption of new models highlights its position in the rapid iteration cycle of the current AI landscape.

Questions people ask about Avian

What hardware does Avian use?

Avian runs its inference services on Nvidia B200 and B300 Blackwell graphics processing units. For dedicated options, it provides eight-GPU HGX nodes with full NVLink domains for high-throughput workloads.

How does Avian charge for its services?

For the API, Avian charges per token consumed, starting at $0.1275 per million tokens with no subscription. For dedicated nodes, customers prepay for terms between 12 months and three years at a locked-in rate.

Is my data stored when I use Avian?

No. Avian states that it has a zero data retention policy. Prompts and completions are not stored after processing, and the infrastructure is GDPR and CCPA compliant.

Can I use Avian with my existing code?

Yes. The API is compatible with the OpenAI standard, allowing you to switch to Avian by changing just one line of code. It works with tools like Cursor and Claude Code.

Good reading and sources

    Where this page comes from

    There is no English Wikipedia article on Avian yet, so the facts on this page come from the sources listed above: avian.io/about, avian.io/. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryNeocloud wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.

    It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

    The weekly digest

    Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.