Not established from the available sources.
META · MODEL RELEASE TRACKER
Llama 4 Maverick
Llama 4 Maverick is a 400-billion-parameter model from Meta.CURRENT SNAPSHOT1/5 DIMENSIONS WITH DATA
The dimensions that change the decision.
Not established from the available sources.
Not established from the available sources.
Proof generation speed with FlashProve on H100Approximately 110 tokens per second on a single H100 GPU (measured for Llama 4 Maverick)
Proof size and verification time with FlashProveEach proof is about 19 megabytes and can be verified in 17 seconds on a standard CPU (for Llama 4 Maverick)
Inference overhead when using FlashProveAdds overhead of roughly 70× to 95× compared to normal inference (for Llama 4 Maverick)
Not established from the available sources.
VERIFIABLE FACTS
Every value stays attached to a source and date.
CAPABILITIES · Parameter countDEVELOPER CLAIM
400 billion parameters
BENCHMARKS · Proof generation speed with FlashProve on H100DEVELOPER CLAIM
Approximately 110 tokens per second on a single H100 GPU (measured for Llama 4 Maverick)
BENCHMARKS · Proof size and verification time with FlashProveDEVELOPER CLAIM
Each proof is about 19 megabytes and can be verified in 17 seconds on a standard CPU (for Llama 4 Maverick)
BENCHMARKS · Inference overhead when using FlashProveDEVELOPER CLAIM
Adds overhead of roughly 70× to 95× compared to normal inference (for Llama 4 Maverick)
WHAT CHANGED