deepinfra
DeepInfra
37 models on DeepInfra, with the price per million tokens and who ends up holding your prompt.
- Company based in
- 🇺🇸 United States
- How it serves
- Runs 24 models itself · forwards 13 models to their makers
- Training on your prompts (models it runs)
- no training
- Data policy source
- docs.deepinfra.com
- Policy checked
- 2026-10-22
108 model retirements from DeepInfra →
Models
| Model | Input $/M | Output $/M | Blended* | Holds your prompt | Training |
|---|---|---|---|---|---|
| Claude Fable 5 | $10.00 | $50.00 | $20.00 | forwards 🇺🇸 United States | off by default |
| Claude Haiku 4.5 | $1.00 | $5.00 | $2.00 | forwards 🇺🇸 United States | off by default |
| Claude Opus 4.7 | $5.00 | $25.00 | $10.00 | forwards 🇺🇸 United States | off by default |
| Claude Opus 4.8 | $5.00 | $25.00 | $10.00 | forwards 🇺🇸 United States | off by default |
| Claude Opus 5 | $5.00 | $25.00 | $10.00 | forwards 🇺🇸 United States | off by default |
| Claude Sonnet 4.6 | $3.00 | $15.00 | $6.00 | forwards 🇺🇸 United States | off by default |
| Claude Sonnet 5 | $3.00 | $15.00 | $6.00 | forwards 🇺🇸 United States | off by default |
| DeepSeek V4 Flash | $0.09 | $0.18 | $0.11 | runs it 🇺🇸 United States | no training |
| DeepSeek V4 Flash 0731 | $0.06 | $0.18 | $0.09 | runs it 🇺🇸 United States | no training |
| DeepSeek V4 Flash Vision Exp | $0.22 | $0.65 | $0.32 | runs it 🇺🇸 United States | no training |
| DeepSeek V4 Pro | $1.30 | $2.60 | $1.63 | runs it 🇺🇸 United States | no training |
| DeepSeek V4 Pro 0813 | $1.30 | $2.60 | $1.63 | runs it 🇺🇸 United States | no training |
| DeepSeek V4.1 Flash | $0.14 | $0.42 | $0.21 | runs it 🇺🇸 United States | no training |
| DeepSeek V4.1 Flash 1017 | $0.10 | $0.29 | $0.15 | runs it 🇺🇸 United States | no training |
| Gemini 2.5 Pro | $1.25 | $10.00 | $3.44 | forwards 🇺🇸 United States | off by default |
| Gemini 3.1 Flash Lite | $0.25 | $1.50 | $0.56 | forwards 🇺🇸 United States | off by default |
| Gemini 3.5 Flash | $1.50 | $9.00 | $3.38 | forwards 🇺🇸 United States | off by default |
| Gemma 4 31B IT | $0.20 | $0.40 | $0.25 | runs it 🇺🇸 United States | no training |
| Gemma 4 31B IT | $0.27 | $0.76 | $0.39 | runs it 🇺🇸 United States | no training |
| Gemma 4 31B IT | $0.09 | $0.34 | $0.15 | runs it 🇺🇸 United States | no training |
| GLM-5.2 | $0.56 | $1.80 | $0.87 | runs it 🇺🇸 United States | no training |
| GLM-5.3 | $0.56 | $2.50 | $1.05 | runs it 🇺🇸 United States | no training |
| GLM-5.3-Flash | $0.15 | $0.50 | $0.24 | runs it 🇺🇸 United States | no training |
| Kimi K2.6 | $0.75 | $3.50 | $1.44 | runs it 🇺🇸 United States | no training |
| Kimi K3 | $2.85 | $14.25 | $5.70 | runs it 🇺🇸 United States | no training |
| Llama-3.1-70B-Instruct | $0.40 | $0.40 | $0.40 | runs it 🇺🇸 United States | no training |
| Llama-3.1-8B-Instruct | $0.02 | $0.04 | $0.03 | runs it 🇺🇸 United States | no training |
| Llama-3.3-70B-Instruct turbo | $0.10 | $0.32 | $0.16 | runs it 🇺🇸 United States | no training |
| Llama 4 Scout 17B 16E Instruct | $0.10 | $0.30 | $0.15 | runs it 🇺🇸 United States | no training |
| MiMo-V2.6-Flash | $0.09 | $0.18 | $0.11 | runs it 🇺🇸 United States | no training |
| MiMo-V2.6-Pro | $0.39 | $0.78 | $0.49 | runs it 🇺🇸 United States | no training |
| MiniMax-M3 | $0.28 | $1.10 | $0.49 | runs it 🇺🇸 United States | no training |
| Nemotron 3 Super 120B A12B | $0.09 | $0.40 | $0.16 | runs it 🇺🇸 United States | no training |
| Nemotron 3 Ultra 550B A55B | $0.50 | $2.20 | $0.93 | runs it 🇺🇸 United States | no training |
| Nemotron 3.5 Lightning 30B A3B | $0.06 | $0.16 | $0.08 | runs it 🇺🇸 United States | no training |
| Qwen3.7 Max | $2.50 | $7.50 | $3.75 | forwards 🇨🇳 China | no training |
| Qwen3.8 Flash | $0.11 | $0.38 | $0.18 | forwards 🇨🇳 China | no training |
| Qwen3.8 Max | $1.65 | $4.95 | $2.48 | forwards 🇨🇳 China | no training |
| Step 5 Preview | $1.00 | $2.70 | $1.43 | runs it 🇺🇸 United States | no training |
Blended is our calculation: (3 × input + output) ÷ 4, US dollars per million tokens — most calls read far more than they write. Tiers that are cheaper because they train on your data are not used for the “from” price.