> DeepInfra
DeepInfra is an OpenAI-compatible inference cloud for open-weight models, with per-token rates published on a single pricing page. Examples from that page: DeepSeek-V4-Flash-0731 at $0.06 per 1M input tokens ($0.015 cached) and $0.18 per 1M output; Qwen3.5-9B at $0.10/$0.15; DeepSeek-V3.2 at $0.26/$0.38; Kimi-K3 at $2.85 per 1M input and $14.25 per 1M output; embeddings from $0.005 per 1M input tokens. Models without fixed token rates are billed on execution time, and custom LLM deployments are billed per minute of GPU uptime across A100 ($0.89/GPU-hour) through B300 ($4.89/GPU-hour). There is no free credit: the pricing page states you must add a card or pre-pay before you can use the service.
// pros
- Cached-input pricing is published per model and is a fraction of standard input (DeepSeek-V4-Flash-0731 at $0.015 per 1M cached tokens against $0.06 standard, Qwen3.5-397B-A17B at $0.22 against $0.45).
- Three request service tiers are documented on the pricing page: Standard at 1x base price, Priority at 1.5x for faster time-to-first-token, and Flex at 0.8x for asynchronous non-production work, enabled per request via service_tier.
- Embeddings are billed from $0.005 per 1M input tokens on models such as bge-base-en-v1.5 and all-MiniLM-L6-v2, and reranking shares the account, which makes RAG stacks cheap to run alongside chat.
- Drop-in migration: the docs state you can point the OpenAI SDK at https://api.deepinfra.com/v1/openai and keep existing code, with 100+ LLMs available.
- Dedicated capacity is priced per GPU-hour with per-minute billing and weekly invoicing (A100 80GB $0.89, H100 80GB $2.20, H200 141GB $2.69, B200 180GB $3.69, B300 270GB $4.89).
// cons
- No free tier and no trial credit: the pricing page states you have to add a card or pre-pay or you won't be able to use the service.
- Each account is limited to 200 concurrent requests, and lifting that ceiling requires contacting DeepInfra directly.
- Output pricing at the top of the catalogue is high relative to input (Kimi-K3 at $14.25 per 1M output tokens), so agentic or long-generation workloads get expensive quickly.
- Non-token models are billed for inference execution time, and custom LLM deployments are invoiced weekly in minute granularity, so idle dedicated capacity is chargeable.
DeepInfra fits token-metered text, embedding and reranking workloads where you want published per-1M rates and cached-input discounts; pair it with a media-generation provider such as fal.ai, since image and video on DeepInfra are billed by execution time rather than by output unit.
written by Mike Fleming, Uprouter editorial · verified against official sources September 16, 2026
Free tier present — value not quantified
This provider offers a free tier, but its terms don’t convert cleanly into a USD figure (e.g. request-count limits or capacity-dependent pools). We refuse to print a misleading zero here. See the note below and the official pricing page for the exact limits.
note: No free tier stated on official pages (checked 2026-09-16): deepinfra.com/pricing states that you must add a card or pre-pay before using the service, with no starter credit offered.
| Plan | Type | Monthly | Input /1M | Output /1M | Note |
|---|---|---|---|---|---|
| Pay-as-you-go | payg | — | Free | Free | Cheapest tracked model rate (OpenRouter-normalized). |
| DeepSeek-V4-Flash | payg | — | $0.090 | $0.180 | Base price for standard tier. |
| Llama-3.3-70B-Turbo | payg | — | $0.100 | $0.320 | Optimized Turbo variant. |
Pricing normalized from public sources — always verify with the provider.
// model prices
| Model | Input /1M | Output /1M | Context |
|---|---|---|---|
| Meta: Llama 3.1 405B Instruct | Free | Free | 131k |
| Meta-Llama-3.1-70B-Instruct | Free | Free | 131k |
| Llama-3.1-8B-Instruct | Free | Free | 131k |
| Qwen2.5 72B Instruct | Free | Free | 33k |
| DeepSeek Chat | Free | $0.0000 | 164k |
| Mixtral 8x22B Instruct | $0.0000 | $0.0000 | 66k |
$ Does DeepInfra have a free tier?
Yes — DeepInfra offers a free tier. Always confirm current limits on DeepInfra's official pricing page — free tiers change without notice.
$ How much does DeepInfra cost per 1M tokens?
The cheapest model we track at DeepInfra is Free per 1M input tokens and Free per 1M output tokens. This is normalized from DeepInfra's published pricing — verify with the provider before purchasing, since prices change frequently.
$ Is DeepInfra safe to use?
Uprouter rates DeepInfra as low risk. Risk is moderate-low. Costs are per token with no minimums or seat fees, a spending limit can be set to cap exposure, and prepaid accounts simply draw down their balance, which limits runaway billing. The main constraints are commercial rather than legal: no free tier at all, a hard 200 concurrent-request ceiling per account, and weekly invoicing with per-minute billing for custom deployments. No data-retention or training-use statement appears on the pricing page fetched, so that has to be verified in DeepInfra's terms before handling sensitive data. Risk ratings are editorial, evidence-based, and never influenced by affiliate relationships.
Raw community sentiment, shown unblended. Votes never feed into objective data fields like price, risk, or status — and never into sort order.
No published reviews yet — be the first.
// more_informationprice history · live status · risk & confidence · code snippets
price_history
No recorded changes yet.
live_status
current: unknownNo probes recorded yet.
Uprouter runs its own probes; best-effort, not a guarantee. Probe results depend on our network location and cadence — for production, always use the provider’s own status page.
risk_&_confidence
Risk is moderate-low. Costs are per token with no minimums or seat fees, a spending limit can be set to cap exposure, and prepaid accounts simply draw down their balance, which limits runaway billing. The main constraints are commercial rather than legal: no free tier at all, a hard 200 concurrent-request ceiling per account, and weekly invoicing with per-minute billing for custom deployments. No data-retention or training-use statement appears on the pricing page fetched, so that has to be verified in DeepInfra's terms before handling sensitive data.
Confidence reflects how much verifiable evidence (official docs, probe history, community reports) backs this entry. Risk is our editorial assessment of operator reliability and terms — not financial advice.
code_snippets
# DeepInfra's own API — get a key from their dashboard
curl https://deepinfra.com/chat/completions \
-H "Authorization: Bearer THEIR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "meta-llama/llama-3.1-405b-instruct",
"messages": [{ "role": "user", "content": "Hello" }]
}'Direct to the provider — Uprouter receives nothing; verify their current pricing before purchasing.
Ranked by neutral similarity signals only — live status, free-tier shape and free-credit proximity. Commercial relationships never influence this list.