Model

Gemini 2.5 Flash

CLOSED

Google · Gemini 2.5 family · released May 20, 2025

High-throughput hybrid-reasoning workhorse with a controllable thinking budget.

ReasoningCodingVisionFunction callingTool useAgentic

1637.3

Elo · rank #94

Parameters

Undisclosed

Active params

Undisclosed

Context

1M tokens

Architecture

Sparse Mixture-of-Experts Transformer (natively multimodal)

License

Proprietary (Google API Terms)

Languages

—

API price (in/out)

$0.3 / $2.5

Modalities

text · vision · audio · video

Benchmark results

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

Aider PolyglotCoding61.9%#16

best: Claude Opus 4.5 · 89.4%

AIMEMath72.0%#99

best: GPT-5.2 · 100.0%

FACTS GroundingKnowledge85.3%#2

best: Gemini 2.5 Flash-Lite (Thinking) · 86.8%

GPQA DiamondReasoning82.8%#66

best: GPT-5.6 · 94.6%

Humanity's Last ExamReasoning11.0%#85

best: Claude Sonnet 5 · 57.4%

LiveCodeBenchCoding63.9%#88

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MMLUKnowledge88.4%#24

best: OpenAI o3 · 92.9%

MMMUVision79.7%#25

best: Claude Fable 5 · 89.3%

SimpleQAKnowledge26.9%#17

best: GPT-4.5 · 62.5%

SWE-bench VerifiedCoding60.4%#79

best: Claude Fable 5 · 95.0%

tau-benchAgents47.2%#24

best: Claude Opus 4.1 · 82.4%

Run it locally

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

Input / M tok

$0.3

Output / M tok

$2.5

Gemini 2.5 family

Elo progression across releases

Gemini 2.5 ProMar 20251865.2 Gemini 2.5 FlashMay 20251637.3 Gemini 2.5 Flash-Lite (Non-thinking)Jun 20251193.2 Gemini 2.5 Flash-Lite (Thinking)Jun 20251281.7

Resources

DOCSai.google.dev/gemini-api/docs/models/gemini APIAvailable via Google API

API price $0.3/$2.5 · each benchmark row carries its own source badge (see methodology)