Model

Gemini 1.5 Pro

CLOSED

Google · Gemini 1.5 family · released Feb 15, 2024

Breakthrough 1M-token context (up to 2M for some users); all 1.5 models shut down on Gemini API by Sept 24 2025.

ReasoningCodingVisionFunction callingTool useAgentic

1062.3

Elo · rank #202

Parameters

Undisclosed

Active params

Undisclosed

Context

1M tokens

Architecture

Sparse Mixture-of-Experts Transformer

License

Proprietary (Google API/Vertex AI)

Languages

—

API price (in/out)

No hosted API

Modalities

text · vision · audio · video

Benchmark results

Bar shows position within the tracked field; marker = field best

AI2DVision94.4%#8

best: Molmo 72B · 96.3%

AMCMath46.4%#4

best: Falcon-H1 34B · 69.4%

BIG-Bench HardReasoning89.2%#7

best: ERNIE 4.5 300B-A47B · 94.3%

ChartQAVision87.2%#14

best: MiniMax-VL-01 · 91.7%

DocVQAVision93.1%#24

best: Qwen2-VL-72B · 96.5%

DROPReasoning74.9#37

best: Hunyuan-T1 · 93.1

GPQA DiamondReasoning46.2%#201

best: GPT-5.6 · 94.6%

GSM8KMath90.8%#37

best: Llama 3.1 405B · 96.8%

HellaSwagReasoning93.3%#4

best: Claude 3 Opus · 95.4%

HumanEvalCoding84.1%#45

best: Claude Opus 4.5 · 99.4%

LogicKorHuman preference9.29#2

best: GPT-4o · 9.33

MATH-500Math67.7%#102

best: GPT-5 · 99.4%

MathVistaVision63.9%#36

best: Seed 2.1 Pro · 90.7%

MGSMMath87.5%#18

best: OpenAI o4-mini · 93.7%

MiniF2F (valid)Math29.9%#2

best: DeepSeek-Prover-V2 671B · 90.6%

MMLUKnowledge85.9%#49

best: OpenAI o3 · 92.9%

MMMUVision62.2%#74

best: Claude Fable 5 · 89.3%

RakudaHuman preference9.2#2

best: PLaMo-100B · 9.73

StreamingBenchVision70.3%#1

best: this model · 70.3%

TextVQAVision78.7%#15

best: Molmo 2 8B · 85.7%

VQAv2Vision80.2%#7

best: Reka Edge (2026) · 88.4%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

—

Output / M tok

—

Gemini 1.5 family

Elo progression across releases

Gemini 1.5 ProFeb 20241062.3 Gemini 1.5 FlashMay 2024865.5 Gemini 1.5 Flash-8BOct 2024532.6

Resources

DOCSstorage.googleapis.com/deepmind-media/gemini/gemini_v1_5_report.pdf

API price — · each benchmark row carries its own source badge (see methodology)