Model · Google

Google Gemini Pro 3.1 (preview)

2 appearances · meta-score 58.0/100 · last seen 2026-07-03 · scored under rubric v15

Pooled composite across 2 primary runs: 52.5 ± 10.8 (95% CI) provisional: n < 3, treat as unsettled

$39.67Total cost
32,216,705Tokens
51.2%Cache hit
773Requests
1.29%Failure rate
1.07Tools/response
812,021Tokens per $

Agent role scores

Per-agent-role scores for Google Gemini Pro 3.1 (preview)
Agent role	Score / 10
Test Agent	7.20
Sprint-Plan Agent	6.80
Pseudocode Agent	5.96
Spec Agent	5.81
Code-Review Agent	5.76
Code Agent	5.75
Sprint-Review Agent	5.68
Develop Agent	5.68
Orchestrator	5.24
Architect	5.21
Build Agent	4.76
Platform Agent	4.59

Run history (primary model)

Runs where Google Gemini Pro 3.1 (preview) was the primary model, oldest first
Run	SoW	Date	Composite
Gemini Pro 3.1 (preview)	circles	Jun 2026	51.7
Gemini Pro 3.1 (preview)	circles	Jul 2026	53.4

Every run on Favur Evals is scored by the same deterministic engine, on the same Statements of Work. The benchmark is self-funded — no vendor sponsorship, credits, or grants.