DeepInfra
Choose model and scheduling tier together rather than treating the lowest token rate as a production guarantee.
- Starting price
- Example USD 0.10 in / 0.32 out per 1M
- Free access
- No general free quota verified
- Billing unit
- Tokens; tier multipliers
- Platform
- Hosted APIs and dedicated GPUs
- Verified
- 22 Oct 2026
Overview
What you provide
- Send prompts to the exact supported model and specify a service tier only when available.
What you get
- The endpoint returns model output with usage suitable for token-based cost accounting.
Choosing this tool
DeepInfra is worth comparing for applications with a known model family and measurable input/output mix. Flex may suit asynchronous jobs that tolerate occasional unavailability, while Priority costs more to improve scheduling. Together AI is a direct-host comparison; OpenRouter adds provider selection when a single host is insufficient.
A decision worked through
For the quoted Llama model, 10 million input and 2 million output tokens cost USD 1 + USD 0.64 = USD 1.64 at Standard. If the same usage is eligible for Priority, the published 1.5x multiplier gives USD 2.46. The arithmetic assumes no cached-input discount.
DeepInfra alternatives for low-unit-cost inference with service tiers
Compare DeepInfra with Together AI, OpenRouter for low-unit-cost inference with service tiers. Match billing units and operating responsibility before comparing rates.
| Product | Price / unit | Free allowance | Input → output | Key limit |
|---|---|---|---|---|
| DeepInfraRead profileOfficial source ↗ | Llama 3.3 70B USD 0.10/0.32 per 1M | Free quota unverified | Prompts become metered model output. | Priority 1.5x; Flex 0.8x. |
| Together AIRead profileOfficial source ↗ | gpt-oss-120B USD 0.15/0.60 per 1M | Free allowance unverified | Model requests become text or media. | Dedicated services price separately. |
| OpenRouterRead profileOfficial source ↗ | Standard: model cost + 5.5% | 50 free-plan requests/day | Messages become routed responses. | Provider policies still differ. |
Setup & workflow
- Choose a model and confirm its current context and rate table.
- Evaluate Standard with recorded input and output usage.
- Test tier changes under load and apply a bounded retry policy.
Requirements & installation
- Configure API credentials and verify the exact model and service-tier availability.
Limits & review
- Flex can be slower or unavailable, and Priority does not change model capability. Dedicated hardware has a different cost basis from serverless token billing.
Your part
- Validate response quality and retry behavior before routing live workloads to a cheaper tier.
Sources & verification4
This profile is based on official sources, not a hands-on product test.
Vendor descriptions and demos document advertised features. Editorial guidance is based on these sources.
- Official DeepInfra functionalitydeepinfra.com
- Official DeepInfra pricingdeepinfra.com
- Together AI official referencewww.together.ai
- OpenRouter official referenceopenrouter.ai
