Atlas

Models

← All models

Gemini 2.5 Pro Experimental 03-25

Google · 2025-03-25

The March 25, 2025 experimental checkpoint of Gemini 2.5 Pro, kept distinct from preview aliases and the stable June release.

Benchmark scores

BenchmarkScore
AI2D (OpenVLM)89.5
Aider Polyglot68.6
AIME 2024 (avg@8)87.5
AIME 2024-202585.8
AIME 202586.7
AIME 2025 (avg@8)84.2
BALROG43.3
BALROG — BabaIsAI (LLM)56.7
BALROG — BabaIsAI (VLM)53.3
BALROG — BabyAI (LLM)80.0
BALROG — BabyAI (VLM)74.0
BALROG — Crafter (LLM)55.0
BALROG — Crafter (VLM)37.3
BALROG — MiniHack (LLM)17.5
BALROG — MiniHack (VLM)12.5
BALROG — NetHack (LLM)1.7
BALROG — NetHack (VLM)1.6
BALROG — TextWorld (LLM)49.2
BALROG (VLM)35.7
BigCodeBench-Hard Complete36.5
BigCodeBench-Hard Instruct29.7
Capability145.5
CorpFin v259.8
CorpFin v2 - Exact Pages59.9
CorpFin v2 - Max Fitting Context57.8
CorpFin v2 - Shared Max Context61.8
Creative Writing v286.2
Creative Writing v3 Elo1396
Creative Writing v3 Rubric Score16.0
GeoBench3999
Loading Atlas data…