Large Language ModelText Generation

Llama 3.3 70B

Meta's efficient 70B model matching 405B performance

Released 2024-12-06

Context Window

128K

API Access

Available

Benchmark Performance

reasoning

reasoning

Massive Multitask Language Understanding

86.0%

coding

coding

HumanEval

88.4%

math

math

MATH

77.0%

math

Grade School Math 8K

93.0%

conversation

conversation

Chatbot Arena Elo

1255.0Elo