name = "Gemini 2.5 Pro"
description = "Google's proven reasoning model for coding, math, and multimodal analysis"
family = "gemini-pro"
release_date = "2025-06-17"
last_updated = "2025-06-17"
attachment = true
reasoning = true
temperature = true
tool_call = true
structured_output = true
knowledge = "2025-01"
open_weights = false

[limit]
context = 1_048_576
output = 65_536

[modalities]
input = ["text", "image", "audio", "video", "pdf"]
output = ["text"]

[[benchmarks]]
name = "Aider Polyglot"
score = 83.1
metric = "percent correct"
source = "https://aider.chat/docs/leaderboards/"
date = "2025-06-06"

[[benchmarks]]
name = "Artificial Analysis Coding Index"
score = 32
metric = "index"
source = "https://openrouter.ai/google/gemini-2.5-pro/benchmarks"
date = "2026-06-02"

[[benchmarks]]
name = "SciCode"
score = 42.8
metric = "percent correct"
source = "https://openrouter.ai/google/gemini-2.5-pro/benchmarks"
date = "2026-06-02"

[[benchmarks]]
name = "Terminal-Bench Hard"
score = 26.5
metric = "success rate"
source = "https://openrouter.ai/google/gemini-2.5-pro/benchmarks"
date = "2026-06-02"