Meta · Llama 3.3
Llama 3.3 70B (Cerebras)
2000+ tokens/sec on Cerebras wafer chips.
Overall
81
Coding
80
Reasoning
78
Writing
81
Meta ranking
#2 of 4
Discovery ranking
#23 of 118
Recommended for
Ultra low-latencyStreaming apps
Capabilities
Vision inputNo
Image generationNo
Tools / function callingYes
Reasoning modeNo
Open weightsYes
Runs locallyYes