Model Details
Gemini 2.0 Flash
Quick answer: Gemini 2.0 Flash (released February 5, 2025) is Google DeepMind's workhorse model for the agentic era: a 1M-token multimodal model at $0.10/$0.40 per million tokens that added native image generation, computer use (visual actions), and a live multimodal API to the Flash tier for the first time. It was the default model in Google AI Studio at launch.
At a Glance
Where Gemini 2.0 Flash leads
- First Flash-tier model with native image output and computer use (clicking, typing, reading screens)
- Multimodal Live API for real-time streaming audio/video interaction with sub-second latency
- Strong coding: significantly improved over Gemini 1.5 Flash on HumanEval and coding agent tasks
- 1M-token context at $0.10/$0.40 — frontier context at sub-cent-per-call pricing for most tasks
Where it lags
- Reasoning depth below Gemini 2.0 Pro and the Gemini 3 family
- Superseded by Gemini 3 Flash for new builds requiring best-in-class performance
Best for: agentic computer use, real-time multimodal applications, production coding assistants, and workflows requiring long-context processing at moderate cost.
What Gemini 2.0 Flash Is
Gemini 2.0 Flash was released as Google's agent-era model. The defining additions vs Gemini 1.5 Flash were computer use (the model can interpret screenshots and output structured UI actions), native image generation (outputs images without a separate image model), and the Multimodal Live API (low-latency streaming interaction across text, audio, and video).
For Benchgen users, Gemini 2.0 Flash is notable as the first broadly available Flash-tier model capable of full computer use agent loops — tasks like navigating web UIs, filling forms, and interacting with desktop applications without requiring a separate vision-action model.
Specifications
| Field | Value |
|---|---|
| Organization | Google DeepMind |
| API identifier | gemini-2.0-flash |
| Context window | 1,000,000 tokens |
| Max output | 8,192 tokens |
| License | Proprietary |
| Release date | February 5, 2025 |
| Modality | Multimodal (text, image, audio, video input; text + image output) |
| Special capabilities | Computer use, native image generation, Multimodal Live API |
Pricing
| Input (per 1M tokens) | Output (per 1M tokens) | |
|---|---|---|
| Google AI | $0.10 | $0.40 |
Image input: $0.0258/image. Source: Google AI pricing.
Public Benchmark Scores
| Benchmark | Score | Source | Date |
|---|---|---|---|
| MMLU | 83.2% | Google — Gemini 2.0 Flash | 2025-02 |
| HumanEval | 83.0% | Google — Gemini 2.0 Flash | 2025-02 |
| GPQA | 62.1% | Google — Gemini 2.0 Flash | 2025-02 |
Scores reported by Google. Not Benchgen measurements.
Gemini 2.0 Flash vs Alternatives
| Model | Context | Price (in/out per 1M) | Computer use |
|---|---|---|---|
| Gemini 2.0 Flash | 1M | $0.10 / $0.40 | Yes |
| Gemini 1.5 Flash | 1M | $0.075 / $0.30 | No |
| Gemini 3 Flash | 1M | TBD | Yes |
| GPT-4o (Nov 2024) | 128K | $2.50 / $10 | Limited |
Gemini 2.0 Flash adds computer use and image generation at 33% more than Gemini 1.5 Flash. GPT-4o costs 25× more per token at 8× less context. (Scores from respective announcements; not Benchgen measurements.)
Related Models & Benchmarks
- Compare: Gemini 2.0 Flash-Lite · Gemini 3 Flash · Gemini 1.5 Pro
- Benchmarks: MMLU-Pro
Frequently Asked Questions
What is Gemini 2.0 Flash?
Gemini 2.0 Flash is Google's agent-era workhorse model, released February 2025. It adds computer use, native image generation, and a real-time multimodal streaming API to the Flash tier, at $0.10/$0.40 per million tokens with 1M-token context.How much does Gemini 2.0 Flash cost?
$0.10 per million input tokens and $0.40 per million output tokens.Specs sourced from Google — Gemini 2.0. Last updated 2026-06-19.
Gemini 2.0 Flash
Gemini 2.0 Flash is a large language model developed by Google.
Benchmark Leaderboards
This model isn’t on any benchmark leaderboard yet.