llm.ing

Gemma 4 31B IT

Googleopen weightsOpenRouter ↗Hugging Face ↗

Largest Gemma 4 instruction model for open, self-hosted chat and reasoning

Benchmarks

BenchmarkCategoryScoreVariantProvenanceSourceObserved
AA Agentic Indexagentic14.4mirroropenrouterJul 23, 2026
AA Coding Indexcoding43.4mirroropenrouterJul 23, 2026
AA Intelligence Indexintelligence29.4mirroropenrouterJul 23, 2026

Includes index data from Artificial Analysis.

Providers

Live operational stats per hosting endpoint, via OpenRouter.

ProviderQuantContext$/1M in · outUptime 24hLatencyThroughput
Cerebrasfp16131K$0.99 · $1.4999.97%
Chutesfp4131K$0.12 · $0.3787.77%
CoreWeavebf16262K$0.12 · $0.3599.48%
Crusoe262K$0.14 · $0.499.55%
DeepInfrafp8262K$0.13 · $0.3897.02%
DeepInfrafp4262K$0.12 · $0.3794.16%
Friendli262K$0.14 · $0.498.87%
ModelRunfp4262K$0.22 · $0.5599.40%
Morphfp4175K$0.14 · $0.497.74%
Novitabf16262K$0.14 · $0.490.44%
OpenInferencebf16262K$0.1 · $0.3599.35%
Parasailfp8262K$0.15 · $0.495.44%
Phala262K$0.15 · $0.4676.27%
SambaNova131K$0.38 · $1.1599.53%
SiliconFlowfp8262K$0.13 · $0.484.62%
Together262K$0.39 · $0.9796.08%
Venicebf16256K$0.12 · $0.3699.36%