Gemini 3.5 Flash

Released

Gemini 3.5 Flash is an AI model released by Google on Tuesday, May 19 2026, 47 days after Gemma 4. It has a 1M token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover BullshitBench v2, Gray Swan IPI, ProgramBench, SWE-Bench Pro, DeepSWE 1.1, MLE-Bench, and 14 more.

Available from

Googleglobalflex$0.75$4.501M
Google AI Studio—flex$0.75$4.501M
Googleglobal—$1.50$9.001M
Google AI Studio——$1.50$9.001M
Googleus—$1.65$9.901M
Googleglobalpriority$2.70$16.201M
Google AI Studio—priority$2.70$16.201M
USD per 1M tokensChecked October 19, 2026Third-party rates from openrouter.ai, not the lab’s own list price.

Benchmarks

Coding

ProgramBenchProgram reconstruction — The AI receives a working program and its documentation, then builds a replacement from scratch without the original source code, internet access or decompilation. The score is the percentage of 200 programs that pass every behavioral test. We record each model's best published mini-SWE-agent result, including higher reasoning efforts where available. Partial test-pass rates and almost-solved programs do not count toward this score. Equal scores share a rank here; the official board also uses partial progress to break ties. Higher is better.via ProgramBench
0%
#11 of 25
SWE-Bench ProAgentic coding — Can the AI fix real bugs in real software? It's handed actual problems from open-source projects and has to write code that genuinely solves them. Higher is better.
55.1%
#28 of 42
DeepSWE 1.1Agentic coding — Artificial Analysis' independent test of deep, agentic software-engineering work — the AI has to plan and carry out substantial coding tasks end to end. (Version 1.1 of the test.) Higher is better.
37%
#39 of 43
MLE-BenchML engineering — Can the AI do the work of a machine-learning engineer? It competes in real Kaggle competitions — building, training, and tuning models end to end — and the score reflects how well it places. Higher is better.
49.7%
#2 of 5

Terminal & CLI

Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better.
76.2%
#25 of 40

Agentic & tool use

MCP AtlasMulti-step tool use — Can the AI chain together many tools and steps to complete one bigger task, rather than doing just a single thing? Higher is better.
83.6%
#5 of 16
ToolathlonGeneral tool use — Tests how well the AI uses everyday real-world tools and apps to get things done. Higher is better.
56.5%
#2 of 11
BU BenchBrowser agent — Can the AI drive a real web browser to finish tasks — clicking, filling forms, and navigating sites the way a person would? Run by Browser Use on their BU Bench task set. Higher is better.
58%
#5 of 7
OSWorld-VerifiedAgentic computer use — Can the AI actually operate a computer — clicking, typing, and using real apps — to finish tasks on its own? Higher is better.
78.4%
#11 of 32

Reasoning & science

Humanity's Last ExamMultidisciplinary reasoning — Humanity's Last Exam — extremely hard expert questions across many subjects, written so you can't just look up the answer. “No tools” means the AI answers on its own. Higher is better.
40.2%no tools
#22 of 52
ARC-AGI-2Abstract reasoning — Puzzle-style tests of abstract reasoning and pattern-finding — the kind of thing people find easy but AIs often struggle with. Higher is better.
72.1%
#20 of 42

Knowledge work

GDPval-AAKnowledge work — Measures how well the AI does economically valuable knowledge work, judged against human experts. Shown as a rating (like a chess Elo) — higher is better.
1656
#6 of 11
GDPval-AA v2Knowledge work — economically valuable knowledge work (v2, re-based Elo)
1349
#21 of 25

Finance

Finance Agent v2Agentic financial analysis — Tests the AI on real financial-analysis work, like digging through reports and making sound decisions. Higher is better.
57.9%
#5 of 13

Multimodal

CharXiv ReasoningChart reasoning — Can the AI read and reason about complex charts and figures, not just text? Higher is better.
84.2%
84.9%with tools
#11 of 28
with tools
#12 of 19
MMMU-ProMultimodal reasoning — A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better.
83.6%
#3 of 29
Blueprint-Bench 2Spatial reasoning — Can the AI reason about space and layout — for example, understanding a floor plan or blueprint? Higher is better.
33.6%
#4 of 8

Long context

MRCR v2 (8-needle)Long context — Tests whether the AI can find specific details buried inside a very long document (around 128k tokens — roughly a long book). Higher is better.
77.3%128k average
26.6%1M pointwise
128k average
#6 of 13
1M pointwise
#3 of 9

Robustness

BullshitBench v2Nonsense detection — Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better.via BullshitBench
20%
#79 of 106
Gray Swan IPIPrompt injection robustness — Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. Gray Swan's indirect prompt injection benchmark measures how often such an attack succeeds when the attacker gets a single try. Lower is better.
14.1%k = 1
54.2%k = 10
60.5%k = 15
k = 1
#12 of 13
k = 10
#12 of 13
k = 15
#14 of 15

Source: ProgramBench, retrieved 11 September 2026. Other sources are identified on the linked benchmark pages.

Compare Gemini 3.5 Flash with

Gemini 3.5 Flash

About

Gemini 3.5 Flash was introduced in May 2026 as Google's general-purpose Flash model for reasoning, coding and practical agent tasks. The release extended the Flash line's emphasis on serving speed into more sustained work with tools and multimodal material. It was offered across Google's developer platforms and consumer experiences, rather than being limited to a research-only checkpoint.

The documented model accepted text, images, video, audio and PDFs, returning text within a roughly one-million-token input limit. Google also supported functions, structured outputs and execution tools around the model. These integrations made it relevant to applications that needed to act on information as well as summarize it. Later Flash updates replaced this generation, so its release-era capabilities should be kept separate from whatever a rolling endpoint serves afterward.

Frequently asked questions

Gemini 3.5 Flash was released by Google on Tuesday, May 19 2026.

All Google releases

34 tracked