Llama 3.1 Instruct 70B: Intelligence, Performance & Price Analysis
Analysis of Meta's Llama 3.1 Instruct 70B and comparison to other AI models across key metrics including quality, price, performance (tokens per second & time to first token), context window & more. Click on any model to compare API providers for that model. For more details including relating to our methodology, see our FAQs.
Meta has launched a newer model, Llama 3.3 70B. We suggest considering this model instead of Llama 3.1 70B. See the following pages for a comparison of Llama 3.3 70B to other models and Llama 3.3 70B API provider benchmarks.
Comparison Summary
Intelligence:
Llama 3.1 70B is of lower quality compared to average, with a MMLU score of 0.676 and a Intelligence Index across evaluations of 26.
Price:Llama 3.1 70B is cheaper compared to average with a price of $0.76 per 1M Tokens (blended 3:1).
Llama 3.1 70B Input token price: $0.76, Output token price: $0.76 per 1M Tokens.
Speed:Llama 3.1 70B Input token price: $0.76, Output token price: $0.76 per 1M Tokens.
Llama 3.1 70B is slower compared to average, with a output speed of 62.3 tokens per second.
Latency:Llama 3.1 70B has a lower latency compared to average, taking 0.40s to receive the first token (TTFT).
Context Window:Llama 3.1 70B has a smaller context windows than average, with a context window of 130k tokens.
Highlights
Intelligence
Artificial Analysis Intelligence Index; Higher is better
Loading chart...
Speed
Output Tokens per Second; Higher is better
Loading chart...
Price
USD per 1M Tokens; Lower is better
Loading chart...
Parallel Queries:
Prompt Length:
Llama 3.1 Instruct 70B Model Details
Comparisons to Llama 3.1 70B
Llama 3.1 70B
o4-mini (high)
GPT-4.1
o3-pro
o3
Llama 4 Maverick
Gemini 2.5 Pro
Gemini 2.5 Flash (Reasoning)
Claude 4 Opus Thinking
Claude 4 Sonnet Thinking
Magistral Small
DeepSeek R1 0528
Grok 4
Nova Premier
Solar Pro 2 (Reasoning)
MiniMax M1 80k
Llama Nemotron Super 49B v1.5 (Reasoning)
Kimi K2
EXAONE 4.0 32B (Reasoning)
GLM-4.5
Qwen3 235B 2507 (Reasoning)
GPT-4o (Nov '24)
Further details