DeepInfra

Cost-effective inference for open-source models

Cost efficiencyOpen modelsScalingReliabilityOpenAI-compatible
HQ: San Francisco, CAFounded: 2022deepinfra.com

About DeepInfra

DeepInfra offers affordable inference for popular open-source models with automatic scaling and high availability. They specialize in making large models accessible at low cost, with a straightforward per-token pricing model.

Key Models

Llama 4 MaverickQwen 3.5Gemma 4

Live Performance Data

Intelligence Score
50.0
Total Requests
65
Success Rate
4308.0%
Avg Latency
1010ms
Models
117
Tokens Processed
1.2M

DeepInfra Models on Antbase

ModelScoreRequestsSuccess RateAvg Latencyp95 LatencyCapabilities
Claude Opus 555.0110000.0%4959ms4959ms
Gemini 2.5 Pro50.000.0%400ms1000ms
All MiniLM L6 V250.000.0%400ms1000ms
Gemma 3 12b It50.000.0%400ms1000ms
🔧
Kimi K2.5 Turbo50.000.0%400ms1000ms
DeepSeek R1 0528 Turbo50.000.0%400ms1000ms
🧠
Qwen3 235B A22B Instruct 250750.000.0%400ms1000ms
🔧
Gemini 2.5 Flash50.000.0%400ms1000ms
PaddleOCR VL 0.9B50.000.0%400ms1000ms
Qwen3 14B50.000.0%400ms1000ms
🔧
Qwen3 Next 80B A3B Instruct50.000.0%400ms1000ms
🔧
GLM 4.6V50.000.0%400ms1000ms
DeepSeek R1 052850.000.0%400ms1000ms
🧠
Text2vec Base Chinese50.000.0%400ms1000ms
Fibo Edit50.000.0%400ms1000ms
Qwen2.5 72B Instruct50.000.0%400ms1000ms
🔧
DeepSeek V3.1 Terminus50.000.0%400ms1000ms
🔧
NVIDIA Nemotron Nano 9B V250.000.0%400ms1000ms
Clip ViT B 32 Multilingual V150.000.0%400ms1000ms
Nemotron 3 Nano 30B A3B50.000.0%400ms1000ms
Kimi K2.550.000.0%400ms1000ms
Qwen3 Coder 480B A35B Instruct50.000.0%400ms1000ms
🔧
Qwen3 VL 30B A3B Instruct50.000.0%400ms1000ms
🔧
Llama 3.2 11B Vision Instruct50.000.0%400ms1000ms
👁🔧
DeepSeek V3 032450.000.0%400ms1000ms
🔧
Llama Guard 4 12B50.000.0%400ms1000ms
Expand50.000.0%400ms1000ms
Remove Background50.000.0%400ms1000ms
Llama 3.1 Nemotron 70B Instruct50.000.0%400ms1000ms
🔧
GLM 4.650.000.0%400ms1000ms
Qwen3 VL 235B A22B Instruct50.000.0%400ms1000ms
🔧
Crystal50.000.0%400ms1000ms
Kimi K2 Thinking50.000.0%400ms1000ms
🧠
GLM 550.000.0%400ms1000ms
Meta Llama 3.1 70B Instruct Turbo50.000.0%400ms1000ms
🔧
Gpt Oss 120b Turbo50.000.0%400ms1000ms
🔧
MiniMax M2.550.000.0%400ms1000ms
Claude 3 7 Sonnet Latest50.000.0%400ms1000ms
Gemma 3 4b It50.000.0%400ms1000ms
🔧
Janus Pro 1B50.000.0%400ms1000ms
GLM 4.7 Flash50.000.0%400ms1000ms
Qwen3 Coder 480B A35B Instruct Turbo50.000.0%400ms1000ms
🔧
Meta Llama 3 8B Instruct50.000.0%400ms1000ms
🔧
Qwen3 32B50.000.0%400ms1000ms
🔧
Qwen3 30B A3B50.000.0%400ms1000ms
🔧
Seedream 4.550.000.0%400ms1000ms
Seed 2.0 Mini50.000.0%400ms1000ms
Kimi K2 Instruct 090550.000.0%400ms1000ms
L3.1 70B Euryale V2.250.000.0%400ms1000ms
Qwen Image Max50.000.0%400ms1000ms
🔧
Qwen2.5 VL 32B Instruct50.000.0%400ms1000ms
👁🔧
L3.3 70B Euryale V2.350.000.0%400ms1000ms
NVIDIA Nemotron Nano 12B V2 VL50.000.0%400ms1000ms
MiniMax M2.150.000.0%400ms1000ms
Claude 4 Opus50.000.0%400ms1000ms
All Mpnet Base V250.000.0%400ms1000ms
Mistral Small 24B Instruct 250150.000.0%400ms1000ms
🔧
Gpt Oss 120b50.000.0%400ms1000ms
🔧
Bria 3.2 Vector50.000.0%400ms1000ms
Clip ViT B 3250.000.0%400ms1000ms
Phi 450.000.0%400ms1000ms
Replace Background50.000.0%400ms1000ms
Gemini 1.5 Flash 8b50.000.0%400ms1000ms
DeepSeek V350.000.0%400ms1000ms
🔧
Claude 4 Sonnet50.000.0%400ms1000ms
Creative50.000.0%400ms1000ms
MythoMax L2 13b50.000.0%400ms1000ms
Mixtral 8x7B Instruct V0.150.000.0%400ms1000ms
🔧
DeepSeek V3.250.000.0%400ms1000ms
🔧
GLM 4.750.000.0%400ms1000ms
Blur Background50.000.0%400ms1000ms
P Image50.000.0%400ms1000ms
Qwen3 235B A22B Thinking 250750.000.0%400ms1000ms
🧠🔧
OlmOCR 2 7B 102550.000.0%400ms1000ms
Llama 3.3 70B Instruct Turbo50.000.0%400ms1000ms
🔧
L3 8B Lunaris V1 Turbo50.000.0%400ms1000ms
Qwen3 Max50.000.0%400ms1000ms
🔧
Seed 1.850.000.0%400ms1000ms
Llama 3.3 Nemotron Super 49B V1.550.000.0%400ms1000ms
🔧
P Image Edit50.000.0%400ms1000ms
Gpt Oss 20b50.000.0%400ms1000ms
🔧
Enhance50.000.0%400ms1000ms
Paraphrase MiniLM L6 V250.000.0%400ms1000ms
Qwen3 Max Thinking50.000.0%400ms1000ms
🧠🔧
DeepSeek OCR50.000.0%400ms1000ms
Gen Fill50.000.0%400ms1000ms
Janus Pro 7B50.000.0%400ms1000ms
Hermes 3 Llama 3.1 405B50.000.0%400ms1000ms
🔧
Llama 4 Scout 17B 16E Instruct50.000.0%400ms1000ms
🔧
Multi Qa Mpnet Base Dot V150.000.0%400ms1000ms
Fibo50.000.0%400ms1000ms
Meta Llama 3.1 70B Instruct50.000.0%400ms1000ms
🔧
Erase Foreground50.000.0%400ms1000ms
All MiniLM L12 V250.000.0%400ms1000ms
Meta Llama 3.1 8B Instruct50.000.0%400ms1000ms
🔧
Hermes 3 Llama 3.1 70B50.000.0%400ms1000ms
🔧
Gemini 1.5 Flash50.000.0%400ms1000ms
Gemma 3 27b It50.000.0%400ms1000ms
🔧
Olmo 3.1 32B Instruct50.000.0%400ms1000ms
Erase50.000.0%400ms1000ms
Wan2.6 Image Edit50.000.0%400ms1000ms
Qwen Image Edit50.000.0%400ms1000ms
🔧
Bria 3.250.000.0%400ms1000ms
DeepSeek R1 Distill Llama 70B50.000.0%400ms1000ms
🧠
Qwen Image Edit Max50.000.0%400ms1000ms
🔧
Kimi K2.7 Code50.02510000.0%23728ms53673ms
Meta Llama 3.1 8B Instruct Turbo50.000.0%400ms1000ms
🔧
DeepSeek V3.150.000.0%400ms1000ms
🔧
Seedream 450.000.0%400ms1000ms
Llama 4 Maverick 17B 128E Instruct FP850.000.0%400ms1000ms
🔧
NVIDIA Nemotron 3 Super 120B A12B50.000.0%400ms1000ms
Mistral Small 3.2 24B Instruct 250650.000.0%400ms1000ms
🔧
Wan2.6 T2I50.000.0%400ms1000ms
Mistral Nemo Instruct 240750.000.0%400ms1000ms
🔧
Qwen3.5 0.8B45.0210000.0%41258ms43761ms
🔧
Claude Fable 535.0280.0%2297ms3484ms
Claude Sonnet 4 633.090.0%1071ms1525ms

How to Use DeepInfra on Antbase

All DeepInfra models are available through Antbase's unified OpenAI-compatible API. You can either route directly to a specific DeepInfra model or let ANT's intelligent router automatically select the best model for your task.

python
from openai import OpenAI

client = OpenAI(
    base_url="https://antbase.ai/v1",
    api_key="your-ant-api-key",
)

# Let ANT auto-route (may use DeepInfra if optimal)
response = client.chat.completions.create(
    model="ant:auto",
    messages=[{"role": "user", "content": "Hello!"}],
)

# Or route directly to a DeepInfra model
response = client.chat.completions.create(
    model="deepinfra/llama-4-maverick",
    messages=[{"role": "user", "content": "Hello!"}],
)

Why Use DeepInfra Through Antbase?

  • •No separate account needed — access DeepInfra models instantly through your Antbase API key.
  • •Intelligent routing — ANT automatically selects the best DeepInfra model for each task, or falls over to another provider if DeepInfra is slow or down.
  • •Same token prices — Antbase charges zero markup on DeepInfra token prices. You pay exactly what DeepInfra charges.
  • •Automatic failover — if DeepInfra has an outage, your requests seamlessly route to an equivalent model on another provider.