DeepInfra

LLM 5.4/10 Jun 16, 2026

DeepInfra runs ~77 open-weight models on a pay-what-you-use basis and lets new accounts try the API on a small free allowance with no payment details.

Tokens / month: 5M
Requests / min: 60
Concurrency: 200

DeepInfra scores 5.4/10 for free value in LLM. The main constraint: Reported ~$0.50 of free usage to try the API before any payment details are required; pay-per-use thereafter with no upfront cost or contract.

Category LLM Kind Hosted open-model LLM API Caveats Free allowance is tiny (~$0.50, reported) and finite; sustained use requires adding payment. Open-weight catalog only. Card required Yes Eligibility anyone

Free value score

5.4 / 10

Show the math

Value 4.4/10

value = 0.5 $ one-time

weighted score = 0.7 * 4.4/10

Model Quality 9.2/10

value = 51 artificial_analysis_score

weighted score = 0.2 * 9.2/10

Volume 1.8/10

value = 259740 tok/mo

score = 1.8/10

Request Rate 4.6/10

value = 60 req/min

weighted score = 0.1 * 4.6/10

Concurrency 5.6/10

value = 200 parallel

score = 5.6/10

Access 7/10

value = 7/10

score = 7/10

Key metrics

GLM-5.2 51 Best model

5.0M T/mo Free tokens

60 R/min Requests/min

200 concurrent Concurrency

$0.5 Credit

Models

0 free · 100+ total

Limits may be stale — last re-verified outside this section's freshness window.

Model	Family	AA	Access
GLM-5.2	GLM	51.1	paid
Qwen3.7-Max	Qwen	46	paid
DeepSeek-V4-Pro	DeepSeek	44.3	paid
Kimi-K2.6	Kimi	42.8	paid
MiMo-V2.5-Pro	MiMo	42.2	paid
Kimi-K2.7-Code	Kimi	42	paid
DeepSeek-V4-Flash	DeepSeek	40.3	paid
GLM-5.1	GLM	40.2	paid
GLM-5	GLM	39.5	paid
Kimi-K2.5	Kimi	38.1	paid
GLM-4.7	GLM	33.8	paid
DeepSeek-V3.2	DeepSeek	24.7	paid
Qwen3-Max-Thinking	Qwen	24	paid
gpt-oss-120b	GPT	23.8	paid
GLM-4.6	GLM	23	paid
DeepSeek-V3.1-Terminus	DeepSeek	21.1	paid
gpt-oss-20b	GPT	14.9	paid
Llama-4-Maverick-17B-128E	Llama	14.3	paid
Llama-4-Scout-17B-16E	Llama	10	paid
phi-4	Phi	4.9	paid
DeepSeek-R1-0528	DeepSeek	—	paid
DeepSeek-R1-Distill-Llama-70B	DeepSeek	—	paid
DeepSeek-V3-0324	DeepSeek	—	paid
FLUX-2-pro	FLUX	—	paid
gemma-3-27b-it	Gemma	—	paid
gemma-4-31B-it	Gemma	—	paid
Llama-3.3-70B-Instruct-Turbo	Llama	—	paid
Llama-3.3-Nemotron-Super-49B-v1.5	Llama	—	paid
Meta-Llama-3.1-8B-Instruct	Llama	—	paid
MiMo-V2.5	MiMo	—	paid
MiniMax-M2.5	MiniMax	—	paid
Mistral-Small-3.2-24B-Instruct-2506	Mistral	—	paid
NVIDIA-Nemotron-3-Super-120B-A12B	Nemotron	—	paid
Qwen3-235B-A22B-Instruct-2507	Qwen	—	paid
Qwen3-235B-A22B-Thinking-2507	Qwen	—	paid
Qwen3-32B	Qwen	—	paid
Qwen3-Coder-480B-A35B-Instruct-Turbo	Qwen	—	paid
Qwen3-Next-80B-A3B-Instruct	Qwen	—	paid
Qwen3-VL-235B-A22B-Instruct	Qwen	—	paid
Voxtral-Small-24B-2507	Mistral	—	paid

Showing first 0 of a larger catalog.

Compare all free models across providers →

What you get

Try the API with ~$0.50 free, no payment details needed

~77 open-weight models incl. Llama, Mixtral, Qwen

Pay-what-you-use pricing, no contracts or upfront costs

OpenAI-compatible API

Gotchas

~$0.50 trial credit

New accounts reportedly receive approximately $0.50 of free API usage before any payment is required; this is a one-time finite pool that does not regenerate.
Card required for sustained use

DeepInfra's pricing page states 'You have to add a card or pre-pay or you won't be able to use our services', so all usage beyond any trial credit requires payment on file.

Sources

"You have to add a card or pre-pay or you won't be able to use our services."
https://deepinfra.com/pricing
"We limit each account to 200 concurrent requests. If you want more drop us a line"
https://deepinfra.com/pricing
"deepinfra/ — DeepInfra inference cloud ($5 one-time trial credit, no credit card) ... $5 one-time credit on signup (no credit card) ~5M tokens, expires after 90 days 60 RPM (varies by model)"
https://github.com/apmantza/pi-free
"DeepSeek V4 Pro (Reasoning, Max Effort) scores 52 on the Artificial Analysis Intelligence Index, placing it well above average among comparable models (averaging 30)."
https://artificialanalysis.ai/models/deepseek-v4-pro
"Today we're announcing $107 million in Series B funding to scale DeepInfra's inference cloud and expand our global capacity."
https://deepinfra.com/blog/deepinfra-series-b

deepinfra open-weight pay-per-use finite-credit openai-compatible

Open service Source ← All LLM

Alternatives

Same category · similar quota

GroqCloudGroqCloud · freemium · Fast inference

5.3

Very fast hosted inference for open models, with free account-level limits.

23.8 AA6K/min

May be outdatedChecked 29d ago

Novita AINovita AI · finite_credit · Hosted open-model LLM + GPU API

5.6

Novita AI hosts 100+ open models behind an OpenAI-compatible API and grants new users a small free credit plus separate time-boxed sandbox credits, no card required.

$1.00 one-time*51.1 AA519.5K tok

May be outdatedChecked 28d ago

One-timeDead

ScalewayScaleway · finite_credit · Serverless EU LLM inference API

5.2

Scaleway's EU-sovereign serverless Generative APIs give every new customer 1,000,000 free tokens (plus 60 free audio-transcription minutes) across open models like Qwen, Llama, Mistral and gpt-oss.

$0.90 one-time*33.7 AA1M tok

May be outdatedChecked 28d ago

One-time