Spaces:

---

huggingface

/

HuggingDiscussions

like 95

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 !!

Hello Hugging Face Team,

We would like to add TextCLF as a Hugging Face inference provider.

Provider Details
Organization: TextCLF — https://huggingface.co/textclf-ai
Website: https://textclf.com

TextCLF provides low-cost LLM inference powered by our proprietary TQ 4-bit quantization and CUDA inference stack. A key differentiator of TQ is that it is completely calibration-data-free: it requires no calibration dataset, so the quantization is not optimized around or dependent on the distribution of a particular calibration set. This makes TQ particularly attractive for generalization across domains and workloads beyond a calibration dataset, while achieving fidelity competitive with leading quantization methods.

On Qwen 3.8 27B (WikiText), TQ 4-bit achieves 0.0282 mean KL divergence and 92.4% Top-1% agreement versus the full-precision model, with approximately 4.1× model compression.

Our currently available models and pricing per 1M tokens are:

Llama 3.1 8B Instruct: $0.018 input / $0.038 output
Llama 3.3 70B Instruct: $0.08 input / $0.30 output
Qwen 3.8 27B: $0.30 input / $2.50 output ($0.03 cached input)

Coming soon: DeepSeek V4 Flash 0731 and XiaomiMiMo MiMo V2.5.

We have submitted PRs to both Hugging Face client repositories to add TextCLF provider support:

Python: https://github.com/huggingface/huggingface_hub/pull/3895
JavaScript: https://github.com/huggingface/huggingface.js/pull/2022

Could you please enable the Model Mapping API for our organization and provide guidance on registering TextCLF as an inference provider?

Thanks,
TextCLF Team
Email: contact@textclf.com

Hi! Following up on our provider application for LLM Tech:

• huggingface.js PR: huggingface/huggingface.js#2403
• huggingface_hub PR: huggingface/huggingface_hub#4727 (full test file passing)

Our endpoint has been serving live production traffic since Aug 22 (currently via NanoGPT): 100% uptime, zero failed server requests, and the best measured TTFT among the 12 providers listed for this model there. Live status with measured numbers, refreshed every 5 minutes: llmtech.eu/status

Specs: Qwen3.8-27B (NVFP4 on Blackwell), 262,144-token context, streaming, prompt caching billed at $0.04/M. Zero data retention, EU jurisdiction.

The billing endpoint per the provider spec is implemented and tested. The org is set up (huggingface.co/llmtech); we'll add the Team subscription and register the model mapping as soon as you're ready on your side.

Dear Hugging Face team,

we would like to register Gridmetry as an Inference Provider.

We operate production inference infrastructure in a Tier III+ datacenter in Munich, Germany and expose an OpenAI-compatible API.

Our initial model portfolio includes high-demand open-weight models such as:

• Kimi K3
• DeepSeek V4 Pro 0813
• MiniMax M2.7
• gpt-oss-120B
• Gemma-family models

We are specifically focused on providing high-throughput, cost-efficient dedicated inference capacity for production workloads in Europe.
We would like to add an initial capacity of ~20.18 billion tokens/day

Current benchmark examples include:

gpt-oss-120B: ~713 output tok/s
MiniMax M2.7: ~428 output tok/s / ~690 ms TTFT

Infrastructure is hosted under German/EU jurisdiction with ISO 27001:2022, zero-data-retention support and dedicated production capacity.

Hugging Face organization: Gridmetry GmbH

We would be happy to implement the required billing endpoint, Inference-Id handling, /v1/models pricing metadata and Hugging Face client integrations.

Could you please enable us for provider onboarding / model mappings and let us know the next steps?

Daniel Schade
COO, Gridmetry GmbH
dsc@gridmetry.com

Hi HF team!

We'd like to register Corvex (https://huggingface.co/Corvex) as an inference provider.

Corvex (https://www.corvex.ai) is a US-based AI cloud company; Token Factory is our serverless inference product for open-weight models: OpenAI and Anthropic compatible, with streaming, tool calling, structured outputs and reasoning.

Integration status, per the register-as-a-provider guide:

• Task API: OpenAI and Anthropic compatible chat completions - live at https://api.tokenfactory.corvex.cloud/v1
• Billing endpoint: implemented to spec (nano-USD) - ready to share the URL + auth privately
• Inference-Id header: on every response, including streaming
• /v1/models pricing + context_length: exposed in the documented format
• JS helper PR: https://github.com/huggingface/huggingface.js/pull/2430
• Python helper (huggingface_hub) ready to PR once mappings are live
• Hub org set up; ready for the model-mapping API once our account is enabled server-side

Could you enable our account server-side when you get a chance? Happy to provide anything else required.

Thanks!
Benjamin Shehu
Corvex (https://www.corvex.ai)

Hi Hugging Face team. Boundless would like to onboard as an Inference Provider.

• Name: Boundless
• Proposed provider identifier: boundless
• Hugging Face organization: BoundlessNetworksInc
• API base URL: https://api.inference.boundless.network
• API documentation: https://inference.boundless.network/docs
• Status page: https://status.boundless.network
• API compatibility: OpenAI-compatible /v1/chat/completions, including streaming, tool calling, and JSON-schema structured outputs

We propose launching initially with these mappings in staging:

• Qwen/Qwen3.6-35B-A3B-FP8
• nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4
• zai-org/GLM-5.2-FP8

We have prepared the huggingface.js provider integration and can open the upstream PR once you confirm that boundless is the preferred identifier.

The production service also supports the billing contract:

• Streaming and non-streaming inference responses include a unique Inference-Id.
• POST https://inference.boundless.network/api/provider/billing accepts request IDs and returns their costs
• GET https://api.inference.boundless.network/v1/models exposes model pricing and context length
• Reviewer API keys and the billing endpoint credential can be shared privately

Could you please confirm:

1. Whether boundless is the correct provider identifier.
2. Whether BoundlessNetworksInc can be enabled for provider onboarding and the Model Mapping API.
3. Whether you would like the huggingface.js PR opened before or after server-side enablement.
4. The preferred private channel for reviewer keys, billing credentials, and our SVG provider icon.

Thank you! We’re happy to adjust the initial mappings or integration sequence based on your guidance. 🙏

Hi HF team! (cc @Wauplin @SBrandeis @julien-c @hanouticelina)

We'd like to register Aiqre (https://huggingface.co/Aiqre) as an inference provider.

Who we are: Aiqre (https://aiqre.com) is an EU inference provider, operated by Advanced AI s.r.o., Czech Republic. We serve open-weight models with a focus on models that are in demand but underserved, and we can bring requested models online quickly. Zero data retention: prompts and completions are never stored.

Integration status, per the register-as-a-provider guide:

Task API: strictly OpenAI-compatible chat completions with streaming, tool calling and structured outputs, live at https://api.aiqre.com/v1
/v1/models with pricing and context_length in the documented format: https://api.aiqre.com/hf/v1/models
Inference-Id header on every response, streaming included
Billing endpoint implemented to spec (nano-USD, idempotent, 10k-ID batches) — happy to share the URL and auth privately
JS helper PR: https://github.com/huggingface/huggingface.js/pull/2437
Python helper PR: https://github.com/huggingface/huggingface_hub/pull/4790
Hub org is set up.

Could you enable our account server-side when you get a chance? Happy to provide anything else you need (test keys, SVG icons, a docs page PR).

Thanks!
Tomas
Aiqre (https://aiqre.com)

Hi Hugging Face team! (cc @julien-c @Wauplin @SBrandeis @hanouticelina)

We'd like to register IteraCompute as a Hugging Face Inference Provider.

Who we are. IteraCompute is an Asia-focused AI inference infrastructure provider specializing in cost-efficient serving of newly released open-weight models. Our production OpenAI-compatible API and documentation are live.

Initial conversational model portfolio:

Both models use the conversational task and support OpenAI-compatible Chat Completions, streaming, tool calling, JSON-schema structured outputs, and reasoning. They are served from Asia (China). Prompts and completions are not retained by default.

Current integration status, following the register-as-a-provider guide:

• Production task API: https://api.iteracompute.com/v1
• Public model catalogue: https://api.iteracompute.com/v1/models
• Production health endpoint: https://api.iteracompute.com/health/ready
• A dedicated HF compatibility gateway has been implemented for https://api.iteracompute.com/hf, including Inference-Id on streaming and non-streaming responses, HF-format /v1/models, and an idempotent billing endpoint returning per-request cost in nano-USD. This dedicated route is pending production deployment.
• huggingface.js and huggingface_hub provider helpers are prepared on our integration branches and are ready to be opened as upstream PRs once you confirm the preferred provider identifier and integration sequence.
• Initial model mappings are prepared with staging status.

External production-gateway benchmark (September 1, 2026): one warm-up followed by five measured streaming requests per model, using a 160-token output cap from a single external client. TTFT is measured from request start to the first non-empty content or reasoning delta; output throughput is measured from the first token to stream completion.

Organization and resources:

• Proposed provider identifier: iteracompute
• Hugging Face organization: IteraCompute
• Website: https://iteracompute.com
• API documentation: https://iteracompute.com/docs.html
• Privacy policy: https://iteracompute.com/privacy.html

We understand that a Team or Enterprise plan is required before Model Mapping API activation and will complete that prerequisite as part of onboarding.

Could you please confirm:

1. Whether iteracompute is the preferred provider identifier.
2. Whether the IteraCompute organization can be enabled for provider onboarding and the Model Mapping API once the plan and production-route prerequisites are complete.
3. Whether you would like the huggingface.js PR opened before or after server-side enablement.
4. The preferred private channel for test credentials, billing endpoint authentication, our SVG icon, company information, and commercial/provider payout onboarding.

Thank you! We're happy to adjust the initial mappings or integration sequence based on your guidance.

IteraCompute
partnerships@iteracompute.com