The home of open models, datasets and Spaces.
DeepInfra
Low-cost inference API for open models.
SharkScore
#47 in Model APIs & inference
Why it scores 59.1
DeepInfra has a SharkScore of 59.1/100, ranked #47 of 262 in Model APIs & inference (updated 2026-11-19).
Each part compares DeepInfra with other listings in Model APIs & inference. Evidence covers 87% of the score's weight.
- Adoption
- 32 · 30%
- User reviews
- no data · 25%
- Maintenance
- 100 · 20%
- Trust & transparency
- 65 · 15%
- Momentum
- no data · 10%
Paying never changes this score. How ranking works · method v2026-11-18.2
RankShark review of DeepInfra
Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review
We were not paid for this review.
As of 2026-11-18, DeepInfra is a model inference API and GPU service with per-model usage pricing rather than a stated monthly subscription. [E2, E3]
DeepInfra offers hosted inference for language, vision, embeddings, image and video generation, and speech models, plus private model deployments and GPU rental. Its docs describe an OpenAI-compatible API. [E3]
Best for
Not for
- Buyers who need a stated fixed monthly subscription price; the pricing page describes model-dependent usage pricing. [E2] · pricing page
Key features
- Model inference API. The vendor says its OpenAI-compatible API serves 100+ LLMs and supports swapping the base URL while keeping existing code. [E3] · docs
- Multiple model categories. The vendor lists LLMs, vision, embeddings, rerankers, image and video generation, and speech models. [E3] · docs
- Private deployments. The vendor says customers can deploy their own fine-tuned models on A100, H100, H200, B200, and B300 hardware with autoscaling and a private endpoint. [E3] · docs
- GPU clusters. The vendor says customers can rent B200 or B300 clusters with SSH access. [E3] · docs
- Pay-as-you-go pricing. The vendor says pricing varies by model; language models may be billed per token and most other models by inference execution time. [E2] · pricing page
Pros
Cons
- The pricing page describes different billing methods by model, and the supplied evidence does not state a fixed monthly subscription price. [E2] · pricing page
- Free-tier and trial terms are not stated in the supplied pages. [E1, E2, E3] · home page, pricing page, docs
- No free plan or free trial stated on the pages we checked.
- Whether customer data is used for model training is not stated on the pages we checked.
Pricing
| Plan | Price | Limits |
|---|---|---|
| Model-dependent usage pricing | Varies by model and billing method; examples below are per 1M tokens. | Some language models are billed per token; most other models are billed for inference execution time. |
| DeepSeek-V4-Flash-0731 | $0.06 per 1M input tokens; $0.18 per 1M output tokens | 1024k context. |
| Meta-Llama-3.1-8B-Instruct-Turbo | $0.02 per 1M input tokens; $0.04 per 1M output tokens | 128k context. |
| Kimi-K3 | $2.85 per 1M input tokens; $14.25 per 1M output tokens | 1024k context. |
| Voxtral-Mini-3B-2507 | $0.00100 per minute of audio input | Audio input pricing. |
| On-Demand DGX B300 GPUs | $5.89 per instance-hour | Instance-hour pricing. |
Checked 18 Nov 2026 · source
- Free plan
- Not stated (checked 18 Nov 2026)
- Free trial
- Not stated (checked 18 Nov 2026)
- Trains on customer data
- Not stated (checked 18 Nov 2026)
- Data retention
- The terms say submitted or generated customer data is not retained beyond the time needed to process and return the request, after which it is deleted. The privacy policy separately says personal information is removed 30 days after account deletion. · terms, privacy page
- Data regions
- The privacy policy says data may be transferred, stored, and processed outside the user's jurisdiction, specifically in the United States; the homepage also describes US-based data centers. · privacy page, home page
- Data processing agreement
- The terms say a DPA is made available at trust.deepinfra.com and applies to processing personal data in customer data. The linked trust site returned a 403 in RankShark's fetch, so its contents were not reviewed. · terms, security page
- AI models used
- The vendor lists 200+ models and families including DeepSeek, Kimi, GLM, Qwen, Llama, Gemini, Claude, and Nemotron. [E1, E2] · home page, pricing page
- API
- OpenAI-compatible API; docs give the base URL https://api.deepinfra.com/v1/openai. [E3] · docs
- SDKs
- The docs describe using the OpenAI SDK and include a Python example. [E3] · docs
- Webhooks
- Not stated (checked 18 Nov 2026)
- Platforms
- Not stated (checked 18 Nov 2026)
- Security claims
- The vendor says it has a zero-retention policy for API inputs, outputs, and user data. [E1] · home pageThe vendor says its security measures comply with SOC 2 and ISO 27001 standards. [E4] · privacy pageThe vendor says it operates inference infrastructure in US-based data centers. [E1] · home page
What we measured
- Median homepage response · 30 days
- 322 ms over 8 checks
- Tranco traffic rank · list of the day
- #56,421
How we tested · sources
- Hourly uptime and response-time checks of the product's URL.
- Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
- We did not sign up, pay, or use the product as a customer.
- home page: https://deepinfra.com/ · fetched 18 Nov 2026
- pricing page: https://deepinfra.com/pricing · fetched 18 Nov 2026
- docs: https://docs.deepinfra.com/ · fetched 18 Nov 2026
- privacy page: https://deepinfra.com/privacy · fetched 18 Nov 2026
- terms: https://deepinfra.com/terms · fetched 18 Nov 2026
- security page: https://trust.deepinfra.com/ · fetched 18 Nov 2026 · HTTP 403
- 18 Nov 2026: Plans or prices changed on the pricing page.
- 18 Nov 2026: Plans or prices changed on the pricing page.
- 18 Nov 2026: The data-use wording changed.
Something wrong? Report an error with a public link that shows it; corrections re-run the review.
Top alternatives to DeepInfra
See all- 79.3ToolFreemium · from $9/mo
- 77.5
API for building with Claude models.
ToolPaid - 76.1
Proxy and SDK to call 100+ LLM APIs in the OpenAI format.
ToolOpen source - 75.7
Google Cloud's platform to build with Gemini and other models.
ToolPaid