← LLM 
Cerebras
gemma-4-31b
Measured 2026-08-25
* Conditional
Stalls after the lookup on 38.9% of runs — announces the refund, never issues it.
Score
84
Task done
77.72%
62–92% 95% CI bootstrap over items · n=95 · measured 2026-08-25
Refusal quality
90.0%
names the gap 87% · offers a route 90% · 10 probes x 3 iterations · 2 silent turns scored zero
Fabrication
0%
0 of 10 probes (30 probe runs) · ≤31%
Dead-air
53.3%
122 of 229 turns
Tool silence
94.6%
122 of 129 tool-call turns
Stalled
38.9%
35 of 90 runs
TTFT p50
192ms
192–232 p50–p90
Cost / 1M tok
$1.49
Share this result
Embed the live badge
Markdown
[](https://benchmarks.speko.ai/llm/cerebras-gemma-4-31b)
HTML
<a href="https://benchmarks.speko.ai/llm/cerebras-gemma-4-31b"><img src="https://benchmarks.speko.ai/badge/llm/cerebras-gemma-4-31b.svg" alt="Speko llm rank"></a>
URL
https://benchmarks.speko.ai/llm/cerebras-gemma-4-31b