DeepSeek V3.1 is 1.3× cheaper per million tokens (blended). DeepSeek V3.1 has the larger context window (163,840 vs 8,192 tokens, 20.0×).

DeepSeek V3.1 vs R1 Distill Llama 70B

Live catalog fields. Best-in-row is highlighted.

DeepSeek V3.1open

DeepSeek · deepseek__deepseek-chat-v3.1

R1 Distill Llama 70Bopen

DeepSeek · deepseek__deepseek-r1-distill-llama-70b

Identity

Fielddeepseek__deepseek-chat-v3.1deepseek__deepseek-r1-distill-llama-70b
ProviderDeepSeekDeepSeek
Slugdeepseek__deepseek-chat-v3.1deepseek__deepseek-r1-distill-llama-70b
Statuslivelive
Open weightsYesYes
License
HuggingFacedeepseek-ai/DeepSeek-V3.1deepseek-ai/DeepSeek-R1-Distill-Llama-70B

Quality

Fielddeepseek__deepseek-chat-v3.1deepseek__deepseek-r1-distill-llama-70b
Quality index (0–100)

Cost

Fielddeepseek__deepseek-chat-v3.1deepseek__deepseek-r1-distill-llama-70b
Input $ / 1M tokens$0.25$0.80
Cached input $ / 1M$0.13
Output $ / 1M tokens$0.95$0.80

Context

Fielddeepseek__deepseek-chat-v3.1deepseek__deepseek-r1-distill-llama-70b
Max context tokens163,8408,192
Max output tokens32,7687,372

Modalities

Fielddeepseek__deepseek-chat-v3.1deepseek__deepseek-r1-distill-llama-70b
Inputtexttext
Outputtexttext