Love that I can call DeepSeek R1 directly from the Hub 🔥

from huggingface_hub import InferenceClient

client = InferenceClient(
    provider="together",
    api_key="xxxxxxxxxxxxxxxxxxxxxxxx"
)

messages = [
    {
        "role": "user",
        "content": "What is the capital of France?"
    }
]

completion = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-R1", 
    messages=messages, 
    max_tokens=500
)

print(completion.choices[0].message)

Is it possible to set a monthly payment budget or rate limits for all the external providers? I don't see such options in billings tab. In case a key is or session token is stolen, it can be quite dangerous to my thin wallet:(

@benhaotang you already get spending notifications when crossing important thresholds ($10, $100, $1,000) but we'll add spending limits in the future

@benhaotang you already get spending notifications when crossing important thresholds ($10, $100, $1,000) but we'll add spending limits in the future

Thanks for your quick reply, good to know!

Would be great if you could add Nebius AI Studio to the list :) New inference provider on the market, with the absolute cheapest prices and the highest rate limits...

Could be good to add featherless.ai

TitanML !!

Any inference provider you love, and that you'd like to be able to access directly from the Hub?

Hi! I would be very happy if you would add the NEURALWATT.COM
It is very very cheap
Thanks in advance!!!

Hi @julien-c and the Hugging Face team,

We would like to apply to become a Hugging Face Inference Provider for automatic speech recognition.

Proposed provider ID: vibevoice-asr
Hugging Face organization: https://huggingface.co/vibevoice-asr
Website: https://vibevoice-asr.com
Initial model: microsoft/VibeVoice-ASR
Task: automatic-speech-recognition
Contact: hello@mail.vibevoice-asr.com

We are building a dedicated inference service for VibeVoice-ASR and intend to support Hugging Face Routed Requests.

Implementation is already underway, and we expect to have a staging-ready endpoint, initial pricing, and preliminary performance results available during the week of August 10, 2026.

We are following the official provider registration guide and plan to provide:

• Hugging Face ASR task-compatible input and output handling
• an Inference-Id header for every successful response
• a billing endpoint returning costNanoUsd
• a huggingface.js provider integration, followed by huggingface_hub integration once the Hub integration is operational
• a staging model mapping backed by a publicly reachable endpoint for Hugging Face automated validation
• usage-based pricing calculated from audio duration

We would like to begin the onboarding process early so that we can align our implementation with Hugging Face requirements before our staging endpoint is ready.

We understand that a Team or Enterprise organization plan is required before accessing the Model Mapping API, and we are prepared to upgrade before reaching that stage.

Could you please advise:

1. whether our organization can begin the provider onboarding process while the infrastructure is being completed;
2. who we should coordinate with for server-side enablement, commercial arrangements, and payout onboarding;
3. whether there are any ASR-specific requirements beyond the published provider guide.

We will share our staging endpoint, pricing, performance results, and integration PRs as soon as they are ready.

We are an independent inference provider and are not affiliated with or endorsed by Microsoft.

Thank you!

Dear Hugging Face Team,

I am writing on behalf of Zenlayer to express our strong interest in partnering with Hugging Face as an AI inference provider for your platform.

Zenlayer (https://www.zenlayer.com/) is a leading global edge cloud service provider headquartered in Los Angeles, United States, with offices worldwide. We specialize in delivering ultra-low-latency digital infrastructure and edge computing solutions globally.

By integrating Zenlayer as an AI inference provider, Hugging Face users can benefit from:

• Global Private Backbone Network: Our extensive high-speed dedicated network spans over 300 POPs globally, ensuring secure, high-throughput, and ultra-reliable data transmission.
• Ultra-Low Latency AI Access: By deploying inference workloads closer to end users at the edge, we significantly reduce response times and deliver smooth, real-time AI experiences.
• Scalable Infrastructure: Comprehensive support for high-demand AI workloads across emerging and established markets worldwide.

We would love to discuss potential integration opportunities and explore how we can collaborate to enhance AI deployment for your community.

Please let us know who would be the best point of contact to take this discussion forward.

Best regards,

Zenlayer Team

contact us: zga-admin@zenlayer.com

Hi Hugging Face team,

we'd like to onboard Kosmik Compute as a Hugging Face Inference Provider.

Our initial integration is complete, and we've submitted the huggingface.js provider PR: https://github.com/huggingface/huggingface.js/pull/2378

Kosmik Compute (kosmik)
https://koscompute.com
API: https://api.koscompute.com/v1

We're an EU-based inference provider operating our own GPU infrastructure, with zero data retention, hosted in an ISO-certified data center.

Our initial model is Qwen/Qwen3.8-27B-FP8, served as qwen/qwen3.8-27b, with:

• 262k text / 131k vision context and up to 32k output
• streaming, tools, structured outputs, reasoning and vision
• Inference-Id, HF billing integration and /v1/models pricing/context metadata

We have allocated 4x NVIDIA RTX PRO 6000 GPUs for the HF launch, with capacity to scale to up to 32 GPUs based on demand.

The integration has been validated end-to-end, including streaming, tool calling, strict structured outputs, vision, billing reconciliation and concurrent traffic.

Could you please enable provider ID kosmik server-side so we can create our initial staging model mapping?

We're happy to provide billing credentials or any additional technical or commercial information.

Thanks,
Stanislav Zavadil
Kosmik Compute
stanislav@koscompute.cz
https://koscompute.com

Hi Hugging Face team,

I’m following up on MoonMath.ai’s request to join Hugging Face Inference Providers.

We would like to make Zro, our inference platform, available as a provider through the Hugging Face Hub. Our initial focus is reliable, high-performance text generation and chat inference for open models, served on AMD CDNA3 hardware, including MI300X-class infrastructure.

Zro is built around production inference work in the open ecosystem:
Zro: https://zro.moonmath.ai/
MoonMath: https://www.moonmath.ai/
Our SGLang contributions: https://zro.moonmath.ai/posts/zro-loves-sglang
Our MI300X MLA decode work: https://zro.moonmath.ai/posts/mla-kimi-k2-7-mi300x
GitHub: https://github.com/moonmath-ai/zro-inference

We are particularly interested in integrating with the Hub so users can select Zro as an inference provider for supported models through Hugging Face’s standard inference clients and model pages. We can provide the endpoint, model availability, pricing, billing integration, and any other technical details needed to complete the provider integration.

Could you please point us to the right person or next step for the provider onboarding process? We would also appreciate guidance on which requirements you would like us to complete first.

Thank you,
Otsar Shalmoni
MoonMath.ai / Zro

Hi Hugging Face team,

We would like to begin the onboarding process for VeloInfer as a Hugging Face Inference Provider.

Provider name: VeloInfer
Proposed provider ID: veloinfer
Hugging Face organization: https://huggingface.co/VeloInfer

VeloInfer

VeloInfer is building fast, reliable, and cost-efficient inference infrastructure for open-source AI models.

Initial deployment

Our first release will provide inference for the Qwen3.8 27B model, running on a distributed fleet of:

• NVIDIA GeForce RTX 5090 GPUs
• NVIDIA RTX 6000 GPUs

The service will expose an OpenAI-compatible API with streaming chat completions and usage-based pricing.

Compute infrastructure

VeloInfer is integrating GPU capacity from multiple infrastructure providers, including:

• Vast.ai
• RunPod
• Additional privately operated GPU infrastructure partners

We plan to manage these resources through a unified serving, routing, monitoring, and billing layer to provide consistent API behavior across different GPU environments.

Model roadmap

After the initial Qwen3.8 27B deployment, we plan to expand support to larger models, including:

• Kimi K3
• Qwen3.8 Max
• Additional high-demand open and open-weight language models

Future model availability will depend on model licensing, infrastructure requirements, and technical validation.

Multimodal and generative AI

VeloInfer also plans to support workloads beyond text generation, including multimodal and generative models such as:

• Qwen-Image
• MiniMax H3
• Additional image and multimodal generation models

What we focus on

• Low-latency and high-throughput inference
• OpenAI-compatible APIs
• Streaming responses
• Distributed GPU resource orchestration
• Model routing and scalable serving
• GPU inference optimization
• Transparent usage-based pricing

Hugging Face integration

We are preparing VeloInfer for integration with Hugging Face Inference Providers.

Our public API endpoints, supported model catalog, pricing, performance benchmarks, and documentation will be published as the infrastructure reaches staging readiness.

Status

The initial inference infrastructure and Hugging Face provider integration are currently under development.

Thanks,
Zhihua Song
VeloInfer

RunInfra, request to become an official Inference Provider

Hi HF team, we'd like to register RunInfra as an Inference Provider.

RunInfra (https://runinfra.ai) runs production model APIs at https://api.runinfra.ai/v1: OpenAI-compatible chat completions with streaming and tool calling, per-token billing, prepaid credits, and published per-model pricing. Currently serving DeepSeek V4 Flash, Nemotron 3.5 Lightning 30B, Qwen3.8 2.4T A95B, and Qwen3.8 27B.

Ready on our side:

• Standard conversational task API (OpenAI-compatible chat completions).
• GET /v1/models publishes pricing and context_length per model.
• A billing cost API implementing your per-request cost contract in nano-USD, answerable within your settlement window.
• Provider Hub org on a Team plan: https://huggingface.co/RunInfra, ready for model mappings.
• Provider modules for huggingface.js and huggingface_hub plus the hub-docs provider page are prepared; we'll open the PRs in your preferred order (JS first, per your provider guide).

Happy to provide anything else you need. Contact: jaber@runinfra.ai.