DeepSeek

DeepSeek R1 Distill Llama 70B

Zero-eval

by DeepSeek

+
+
+
+
About

DeepSeek-R1-Distill-Llama-70B was created through knowledge distillation from DeepSeek-R1 into a Llama-based architecture, designed to transfer reasoning capabilities to a widely-used open-source foundation. Built to combine DeepSeek's reasoning innovations with Llama's ecosystem compatibility, it enables broader access to advanced reasoning techniques.

+
+
+
+
Pricing Range
Input (per 1M)$0.10 -$0.10
Output (per 1M)$0.40 -$0.40
Providers1
+
+
+
+
Timeline
AnnouncedJan 20, 2025
ReleasedJan 20, 2025
+
+
+
+
Specifications
Training Tokens14.8T
+
+
+
+
License & Family
License
MIT
Performance Overview
Performance metrics and category breakdown

Overall Performance

4 benchmarks
Average Score
76.0%
Best Score
94.5%
High Performers (80%+)
2

Performance Metrics

Max Context Window
256.0K
Avg Throughput
37.0 tok/s
Avg Latency
1ms
+
+
+
+
All Benchmark Results for DeepSeek R1 Distill Llama 70B
Complete list of benchmark scores with detailed information
MATH-500
text
0.94
94.5%
Self-reported
AIME 2024
text
0.87
86.7%
Self-reported
GPQA
text
0.65
65.2%
Self-reported
LiveCodeBench
text
0.57
57.5%
Self-reported
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+