# Gemini 3.5 Flash

Gemini 3.5 Flash is an AI model released by Google on May 19 2026. At release it scored 56.5% on Toolathlon, 57.9% on Finance Agent v2 and 83.6% on MMMU-Pro.

## Facts

| Field | Value |
| --- | --- |
| Model | Gemini 3.5 Flash |
| Developer | Google |
| Release date | Tuesday, May 19 2026 |
| Licensing | Proprietary |

## API pricing

All rates in USD per 1,000,000 tokens, pay-as-you-go.

| Tier | Input | Cached input | Output |
| --- | --- | --- | --- |
| Standard | $1.50 | $0.15 | $9.00 |

Output prices include reasoning tokens.
Verified August 18, 2026 against the first-party source: https://ai.google.dev/gemini-api/docs/pricing

## Benchmark scores published at release

| Benchmark | Score | What it measures |
| --- | --- | --- |
| BullshitBench v2 | 20% | 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. |
| Gray Swan IPI (k = 1) | 14.1% | 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. |
| Gray Swan IPI (k = 10) | 54.2% | Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 10 tries and counts an attack as successful if any of them works. Lower is better. |
| Gray Swan IPI (k = 15) | 60.5% | Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 15 tries and counts an attack as successful if any of them works. Lower is better. |
| SWE-Bench Pro | 55.1% | 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. |
| CursorBench v3.2 | 48.8% | Cursor's own test of harder, real-world coding tasks inside a code editor, on the refreshed v3.2 task set. Scores aren't comparable with v3.1. Higher is better. |
| CursorBench v3.1 | 49.8% | Cursor's own test of harder, real-world coding tasks inside a code editor. Higher is better. |
| DeepSWE 1.1 | 37% | 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. |
| MLE-Bench | 49.7% | 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. |
| Terminal-Bench 2.1 | 76.2% | Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better. |
| MCP Atlas | 83.6% | Can the AI chain together many tools and steps to complete one bigger task, rather than doing just a single thing? Higher is better. |
| Toolathlon | 56.5% | Tests how well the AI uses everyday real-world tools and apps to get things done. Higher is better. |
| BU Bench | 58% | 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. |
| Humanity's Last Exam (no tools) | 40.2% | 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. |
| ARC-AGI-2 | 72.1% | Puzzle-style tests of abstract reasoning and pattern-finding — the kind of thing people find easy but AIs often struggle with. Higher is better. |
| OSWorld-Verified | 78.4% | Can the AI actually operate a computer — clicking, typing, and using real apps — to finish tasks on its own? Higher is better. |
| Finance Agent v2 | 57.9% | Tests the AI on real financial-analysis work, like digging through reports and making sound decisions. Higher is better. |
| GDPval-AA | 1656 | 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. |
| GDPval-AA v2 | 1349 | economically valuable knowledge work (v2, re-based Elo) |
| CharXiv Reasoning | 84.2% | Can the AI read and reason about complex charts and figures, not just text? Higher is better. |
| MMMU-Pro | 83.6% | A tougher version of MMMU — college-level questions that mix images, diagrams, and text together. Higher is better. |
| Blueprint-Bench 2 | 33.6% | Can the AI reason about space and layout — for example, understanding a floor plan or blueprint? Higher is better. |
| MRCR v2 (8-needle) (128k average) | 77.3% | Tests whether the AI can find specific details buried inside a very long document (around 128k tokens — roughly a long book). Higher is better. |
| MRCR v2 (8-needle) (1M pointwise) | 26.6% | Tests whether the AI can find specific details buried inside an enormous document (around 1 million tokens — many books). Higher is better. |
| Arena Elo (Text) | 1476 | Real people chat with two anonymous AIs side by side and vote for the answer they prefer. Votes become a chess-style Elo rating on arena.ai — it measures which AI people actually like, not test scores. Higher is better. |
| Arena Elo (Code) | 1499 | Like the text arena, but people vote on which AI writes better code. The votes become a chess-style Elo rating on arena.ai. Higher is better. |

## Questions and answers

### When was Gemini 3.5 Flash released?

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

### Who made Gemini 3.5 Flash?

Gemini 3.5 Flash was built by Google. Builds the Gemini family of models through Google DeepMind. Integrates AI across Google products.

### How much does Gemini 3.5 Flash cost?

Gemini 3.5 Flash costs $1.50 per million input tokens and $9.00 per million output tokens through the Google API. Cached input is $0.15 per million tokens. Output prices include reasoning tokens. Rates are pay-as-you-go API prices verified against Google's published pricing on August 18, 2026.

### What benchmark scores did Gemini 3.5 Flash get?

Gemini 3.5 Flash reports 26 tracked benchmark scores — BullshitBench v2: 20%; Gray Swan IPI (k = 1): 14.1%; Gray Swan IPI (k = 10): 54.2%; Gray Swan IPI (k = 15): 60.5%; SWE-Bench Pro: 55.1%; CursorBench v3.2: 48.8%; CursorBench v3.1: 49.8%; DeepSWE 1.1: 37%; MLE-Bench: 49.7%; Terminal-Bench 2.1: 76.2%; MCP Atlas: 83.6%; Toolathlon: 56.5%; BU Bench: 58%; Humanity's Last Exam (no tools): 40.2%; ARC-AGI-2: 72.1%; OSWorld-Verified: 78.4%; Finance Agent v2: 57.9%; GDPval-AA: 1656; GDPval-AA v2: 1349; CharXiv Reasoning: 84.2%; MMMU-Pro: 83.6%; Blueprint-Bench 2: 33.6%; MRCR v2 (8-needle) (128k average): 77.3%; MRCR v2 (8-needle) (1M pointwise): 26.6%; Arena Elo (Text): 1476; Arena Elo (Code): 1499. Scores are the figures published at release by Google. It holds the best score among all models tracked here on Toolathlon, Finance Agent v2 and MMMU-Pro.

### Is Gemini 3.5 Flash open source?

No. Gemini 3.5 Flash is a proprietary model. The weights are not published — it is available only through the provider's own API, apps, or partner platforms.

### What came before and after Gemini 3.5 Flash?

Google's previous tracked release was Gemma 4 on Apr 2 2026, 47 days earlier. It was followed by Gemini 3.6 Flash on Jul 21 2026.


---

Canonical page: https://aireleasetracker.com/model/google/gemini-3.5-flash
Full dataset: https://aireleasetracker.com/llms-full.txt · JSON: https://aireleasetracker.com/models.json
Source: AI Release Tracker (https://aireleasetracker.com). Benchmark scores are the figures published by the releasing lab at launch.