Spaces:

---

huggingface

/

HuggingDiscussions

Like 98

Running

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 !!

Hi Hugging Face team,

we are preparing SovSales Inference, an independent OpenAI-compatible inference provider for open-weight models.

Our provider gateway implements OpenAI-compatible chat completions, streaming, token usage reporting, model metadata, bearer authentication, request IDs, rate limiting and a published privacy/data policy. Production GPU-backed serving is currently being staged and will be validated before any model is marked live.

Provider profile:
https://sovereignsalesswarm.de/inference/

API base:
https://sovereignsalesswarm.de/api/inference/v1

Models:
https://sovereignsalesswarm.de/api/inference/v1/models

We would like to integrate SovSales as a Hugging Face Inference Provider. Could you enable our organization for the provider integration process and confirm the next onboarding steps, including the Team or Enterprise requirement?

Organization:
https://huggingface.co/sovsales

Best,
SovSales Inference
info@sovereignsalesswarm.de

Hi HF team (Cc. @julien-c @Wauplin @sbrandeis )👋

We'd like to register IntelCS-AI as an Inference Provider. We serve
open-weight models at floor prices — our edge is cost: we run efficient
GPU fleets and pass the savings through as some of the lowest per-token
prices on the platform.

What's already built and live (per the register-as-a-provider guide):

• OpenAI-compatible inference endpoint (live, bearer-token auth):
  /v1/chat/completions (streaming + non-streaming). Endpoint URL and
  credentials provided in the private exchange.
• Tool calling + structured output verified end-to-end
  (json_schema response_format) — vLLM 0.9.2 backends.
• /v1/models with the full expected shape — id, object,
  created, owned_by, pricing ($/1M tokens), context_length — so
  the comparison table and :cheapest selection work out of the box.
• Billing endpoint implemented to your spec:
  POST /billing/costs, batched requestIds → integer costNanoUsd,
  idempotent, per-request price snapshot at serve time.
• Inference-Id header (UUID) on every response, including
  streaming.
• Latency: TTFT ~0.9 s non-streaming (well under the 5 s budget);
• Autoscaling fleet with health checks and automatic scale-out.
• Hub org: IntelCS-AI —
  Team plan, org card published.

Initial model (ready to map, task conversational):

The lineup rotates as we add capacity — we'll keep mappings in sync.

What we need from you:

1. Server-side enablement for org IntelCS-AI, provider slug
   intelcs-ai-iaas.
2. Token exchange — we'll provide our bearer token(s) for inference
   • billing calls (we can rotate fresh ones for production).
3. Confirmation of the Inference-Id header name.

Once enabled, we'll register the model mapping via
POST /api/partners/intelcs-ai-iaas/models and open the
huggingface.js + huggingface_hub PRs.

Happy to share anything else you need — thanks!
Brad (IntelCS-AI · intelcs.ai)

Hi Hugging Face team,

We operate a Compute as a Service (CaaS) platform, providing customers with on-demand access to high-performance compute infrastructure, including GPU capacity, without the need to own or manage the underlying hardware.

We are headquartered in the UK and have recently expanded our compute capacity into the US and Canada. We are interested in registering as an inference provider on Hugging Face and exploring how our infrastructure could support inference workloads through the Hugging Face platform.

We would appreciate it if someone from the relevant team could get in touch to discuss the registration process and requirements.

Kind Regards,

Mariami

Mariami.Robakidze@bec.uk

Hey! I'm the founder of Kitani an inference API provider serving open source LLMs. We're interested in becoming an official Hugging Face Inference Provider and routing HF inference traffic through our infrastructure.

We already have an OpenAI compatible API and production inference infrastructure. @Wauplin @julien-c , would either of you be able to point me to the right person to coordinate a provider integration with?

Thanks!

My emails admin@kitani.ai

Hi Hugging Face team,

We would like to register as a Hugging Face Inference Provider.

We operate a GPU-backed OpenAI-compatible inference gateway and would like to make GLM-5.3 available through Hugging Face Inference Providers.

Provider information:

• Proposed provider name: Yuxuan
• Model: zai-org/GLM-5.3 / GLM-5.3
• API Base URL: https://api.chunheaj.top/v1
• Models endpoint: https://api.chunheaj.top/v1/models
• Chat Completions endpoint: https://api.chunheaj.top/v1/chat/completions
• Supported task: conversational
• Supported features: chat completions, SSE streaming, token usage reporting, OpenAI-compatible request/response format
• Initial capacity: conservative concurrency of 4 requests to maintain stable latency and reliability

We can provide private test API credentials for validation and are ready to coordinate on the next steps, including the required provider organization setup, model mappings, billing endpoint, and client integration.

Could someone from the Hugging Face Inference Providers team confirm the next onboarding steps and the correct person to coordinate with?

For private coordination and test credentials, please contact us at: gaoja2026@gmail.com

Thank you.

Hi Hugging Face team,

Apres Dynamics (huggingface.co/ApresDynamics) for embeddings, reranking, classification, and light generation.

We saw the note on PR #2491 about the consolidation phase. Are you currently accepting new Inference Provider partners, and is there a waitlist we can join?

Thanks,
Michael Lewis
@ApresDynamics
michael@apresski.ai

Hi Hugging Face team,

We are working on register as provider for our own models, we have questions about model serving provider registration process, we would appreciate it if someone from the relevant team could get in touch to discuss the registration process and requirements.

We'd like to register BosonAI as an Inference Provider for text-to-speech. BosonAI trained and publishes both models — we're the original model developer, not a third-party inference reseller, so our serving stack is built and tuned specifically for this architecture.

Provider details:

• Name: BosonAI, proposed provider id bosonai
• Hugging Face organization: https://huggingface.co/bosonai
• Website: https://boson.ai

Initial models, task text-to-speech:

• https://huggingface.co/bosonai/higgs-tts-3-4b
• https://huggingface.co/bosonai/higgs-tts-2-3b-base

Both models are already deployed and serving production traffic on our own infrastructure
at https://api.boson.ai/v1/audio/speech.

Current status against the register-as-a-provider guide:

• Task API: live in production; not yet in HF's text-to-speech input/output schema —
  happy to either adapt on our side or provide the mapping for a huggingface.js provider
  helper, whichever you prefer.
• Billing endpoint: implementing now (Inference-Id header + idempotent cost-in-nanoUSD
  endpoint per spec).
• huggingface.js integration: not yet started, pending your guidance on provider id and
  sequencing.

Thank you!

Best,
Bang Liu
@BosonAI
bang@boson.ai