Deepinfra: Models Intelligence, Performance & Price
Analysis of Deepinfra's models across key metrics including quality, price, output speed, latency, context window & more. This analysis is intended to support you in choosing the best model provided by Deepinfra for your use-case. For more details including relating to our methodology, see our FAQs. Models analyzed: Llama 3.3 70B (Turbo, FP8), Llama 3.3 70B, Llama 3.1 405B, Llama 3.2 90B (Vision), Llama 3.1 70B (Turbo, FP8), Llama 3.1 70B, Llama 3.2 11B (Vision), Llama 3.1 8B, Llama 3.2 3B, Llama 3.2 1B, Llama 4 Maverick (FP8), Llama 4 Scout, Gemma 2 9B, Gemma 3 12B, Gemma 3 27B, Gemma 3 4B, Mistral Small 3, Mistral NeMo, Mixtral 8x7B, DeepSeek R1 (Turbo, FP4), DeepSeek R1, DeepSeek R1 Distill Llama 70B, DeepSeek V3 (Mar' 25), DeepSeek R1 Distill Qwen 32B, Phi-4, Llama 3.1 Nemotron 70B, Qwen2.5 72B, Qwen2.5 Coder 32B, QwQ-32B, Llama 3 70B, Llama 3 8B, Mistral 7B, DeepSeek V3 (Dec '24), OpenChat 3.5, and QwQ 32B-Preview.
Link:
Deepinfra Model Comparison Summary
Intelligence:
DeepSeek R1 (Turbo, FP4) and DeepSeek R1 are the highest quality models offered by Deepinfra, followed by 
QwQ-32B, DeepSeek V3 (Mar' 25) & DeepSeek R1 Distill Qwen 32B.Output Speed (tokens/s):
Llama 3.2 3B (146 t/s) and Llama 4 Maverick (FP8) (116 t/s) are the fastest models offered by Deepinfra, followed by Llama 3.2 1B, Llama 3 8B & DeepSeek R1 (Turbo, FP4).Latency (seconds): Llama 3.2 3B (0.19s) and 
Mixtral 8x7B (0.21s) are the lowest latency models offered by Deepinfra, followed by Llama 3 8B, Mistral 7B & Llama 3.2 11B (Vision).Blended Price ($/M tokens): Llama 3.2 1B ($0.01) and Llama 3.2 3B ($0.02) are the cheapest models offered by Deepinfra, followed by 
Gemma 3 4B, Llama 3.1 8B & Mistral 7B.Context Window Size: DeepSeek V3 (Mar' 25) (164k) and Llama 4 Maverick (FP8) (131k) are the largest context window models offered by Deepinfra, followed by Llama 4 Scout, QwQ-32B & Llama 3.3 70B (Turbo, FP8).
DeepSeek R1 (Turbo, FP4) and DeepSeek R1 are the highest quality models offered by Deepinfra, followed by QwQ-32B, DeepSeek V3 (Mar' 25) & DeepSeek R1 Distill Qwen 32B.Output Speed (tokens/s): Llama 3.2 3B (146 t/s) and Llama 4 Maverick (FP8) (116 t/s) are the fastest models offered by Deepinfra, followed by Llama 3.2 1B, Llama 3 8B & DeepSeek R1 (Turbo, FP4).Latency (seconds): Llama 3.2 3B (0.19s) and Mixtral 8x7B (0.21s) are the lowest latency models offered by Deepinfra, followed by Llama 3 8B, Mistral 7B & Llama 3.2 11B (Vision).Blended Price ($/M tokens): Llama 3.2 1B ($0.01) and Llama 3.2 3B ($0.02) are the cheapest models offered by Deepinfra, followed by Gemma 3 4B, Llama 3.1 8B & Mistral 7B.Context Window Size: DeepSeek V3 (Mar' 25) (164k) and Llama 4 Maverick (FP8) (131k) are the largest context window models offered by Deepinfra, followed by Llama 4 Scout, QwQ-32B & Llama 3.3 70B (Turbo, FP8).Highlights
Intelligence
Artificial Analysis Intelligence Index; Higher is better
Speed
Output Tokens per Second; Higher is better
Price
USD per 1M Tokens; Lower is better
Parallel Queries:
Prompt Length:
Features | Model Intelligence | Price | Output tokens/s | Latency | End-to-End Response Time | ||||
|---|---|---|---|---|---|---|---|---|---|
Further Analysis | |||||||||
 | DeepSeek R1 (Turbo, FP4) | 33k | 60 | $1.50 | 101.5 | 0.29 | 28.33 | 23.11 | |
| DeepSeek R1 | 64k | 60 | $0.96 | 20.9 | 0.40 | 136.68 | 112.35 | |
| QwQ-32B | 131k | 58 | $0.14 | 28.7 | 0.26 | 104.52 | 86.83 | |
| DeepSeek V3 (Mar' 25) | 164k | 53 | $0.52 | 11.8 | 0.75 | 43.28 | N/A | |
| DeepSeek R1 Distill Qwen 32B | 128k | 52 | $0.14 | 44.6 | 0.25 | 56.35 | 44.88 | |
| Llama 4 Maverick (FP8) | 131k | 51 | $0.30 | 116.4 | 0.24 | 4.54 | N/A | |
| DeepSeek R1 Distill Llama 70B | 128k | 48 | $0.34 | 31.5 | 0.30 | 79.66 | 63.49 | |
| DeepSeek V3 (Dec '24) | 64k | 46 | $0.59 | 16.8 | 0.52 | 30.27 | N/A | |
| Llama 4 Scout | 131k | 43 | $0.15 | 88.4 | 0.27 | 5.93 | N/A | |
| QwQ 32B-Preview | 33k | 43 | $0.26 | 45.2 | 0.24 | 55.53 | 44.24 | |
| Llama 3.3 70B (Turbo, FP8) | 128k | 41 | $0.20 | 34.0 | 0.26 | 14.95 | N/A | |
| Llama 3.3 70B | 128k | 41 | $0.27 | 19.2 | 0.53 | 26.56 | N/A | |
| Llama 3.1 405B | 33k | 40 | $0.90 | 26.5 | 0.44 | 19.33 | N/A | |
| Qwen2.5 72B | 33k | 40 | $0.27 | 40.0 | 0.29 | 12.79 | N/A | |
|  Phi-4 | 16k | 40 | $0.09 | 35.8 | 0.28 | 14.23 | N/A | |
| Gemma 3 27B | 128k | 38 | $0.07 | 39.7 | 0.35 | 12.94 | N/A | |
|  Llama 3.1 Nemotron 70B | 128k | 37 | $0.27 | 31.5 | 0.30 | 16.17 | N/A | |
| Qwen2.5 Coder 32B | 33k | 36 | $0.10 | 49.3 | 0.25 | 10.40 | N/A | |
| Llama 3.1 70B (Turbo, FP8) | 128k | 35 | $0.20 | 35.3 | 0.30 | 14.44 | N/A | |
| Llama 3.1 70B | 128k | 35 | $0.27 | 34.4 | 0.38 | 14.92 | N/A | |
| Mistral Small 3 | 32k | 35 | $0.09 | 61.8 | 0.26 | 8.36 | N/A | |
| Gemma 3 12B | 128k | 34 | $0.06 | 24.6 | 0.59 | 20.94 | N/A | |
| Llama 3.2 90B (Vision) | 33k | 33 | $0.36 | 38.1 | 0.39 | 13.53 | N/A | |
| Llama 3 70B | 8k | 27 | $0.27 | 32.4 | 0.50 | 15.95 | N/A | |
| Gemma 3 4B | 128k | 24 | $0.03 | 89.1 | 0.31 | 5.92 | N/A | |
| Llama 3.1 8B | 128k | 24 | $0.04 | 43.3 | 0.24 | 11.79 | N/A | |
| Gemma 2 9B | 8k | 22 | $0.04 | 25.1 | 0.58 | 20.51 | N/A | |
| Llama 3 8B | 8k | 21 | $0.04 | 102.5 | 0.21 | 5.09 | N/A | |
| Mistral NeMo | 128k | 20 | $0.06 | 43.7 | 0.27 | 11.72 | N/A | |
| Llama 3.2 3B | 128k | 20 | $0.02 | 145.8 | 0.19 | 3.62 | N/A | |
| Mixtral 8x7B | 33k | 17 | $0.24 | 96.4 | 0.21 | 5.39 | N/A | |
| Mistral 7B | 8k | 10 | $0.04 | 81.8 | 0.21 | 6.32 | N/A | |
| Llama 3.2 1B | 128k | 10 | $0.01 | 114.4 | 0.25 | 4.62 | N/A | |
| Llama 3.2 11B (Vision) | 128k | $0.06 | 45.6 | 0.23 | 11.18 | N/A | ||
|  OpenChat 3.5 | 8k | $0.06 | 48.2 | 0.25 | 10.61 | N/A | ||
Key definitions
Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).
Output Speed: Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).
Latency (Time to First Token): Time to first token received, in seconds, after API request sent. For reasoning models which share reasoning tokens, this will be the first reasoning token. For models which do not support streaming, this represents time to receive the completion.
Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Output Price: Price per token generated by the model (received from the API), represented as USD per million Tokens.
Input Price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.
Time period: Metrics are 'live' and are based on the past 72 hours of measurements, measurements are taken 8 times a day for single requests and 2 times per day for parallel requests.