Google DeepMind

Gemini 3 Flash Preview

Tool callingStructured outputReasoningImage inputAudio inputVideo inputFile input
Gemini 3 Flash Preview launch post cover

New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs

Context
1.0M
Released
10mo ago
Input / 1M tokens
$0.50
Output / 1M tokens
$3.00
Cached input / 1M tokens
$0.05
Cache write / 1M tokens
$0.08

Sources:models.devOpenRouter

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.

Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.

Other published benchmarks

EvaluationReported score and source
ARC-AGI-2
1.3%
Third-party evaluator · gemini-3-flash-preview
External evaluation via Epoch AI
Original-source terms apply
FrontierMath-2025-02-28-Private
35.6%
Epoch AI · gemini-3-flash-preview
Epoch AI evaluation
FrontierMath-Tier-4-2025-07-01-Private
4.2%
Epoch AI · gemini-3-flash-preview
Epoch AI evaluation
FrontierMath-Tier-4-v2-Private
17.1%
Epoch AI · gemini-3-flash-preview
Epoch AI evaluation
FrontierMath-Tiers-1-3-v2-Private
51.2%
Epoch AI · gemini-3-flash-preview
Epoch AI evaluation
GPQA diamond
83.2%
Epoch AI · gemini-3-flash-preview
Epoch AI evaluation
GPQA diamond
89.4%
Epoch AI · gemini-3-flash-preview_high
Reasoning effort: high
Epoch AI evaluation
OTIS Mock AIME 2024-2025
92.8%
Epoch AI · gemini-3-flash-preview
Epoch AI evaluation
OTIS Mock AIME 2024-2025
95.6%
Epoch AI · gemini-3-flash-preview_high
Reasoning effort: high
Epoch AI evaluation
SimpleQA Verified
66.8%
Epoch AI · gemini-3-flash-preview_high
Reasoning effort: high
Epoch AI evaluation
SWE-Bench verified
75.4%
Epoch AI · gemini-3-flash-preview
Epoch AI evaluation
Terminal Bench
51.0%
Terminal-Bench v2 Leaderboard · gemini-3-flash-preview
Agent: Gemini CLI
Terminal-Bench v2 Leaderboard
Original-source terms apply
Terminal Bench
47.4%
https://www.tbench.ai/leaderboard/terminal-bench/2.0 · gemini-3-flash-preview
Agent: Gemini CLI
https://www.tbench.ai/leaderboard/terminal-bench/2.0
Original-source terms apply
Terminal Bench
64.3%
https://www.tbench.ai/leaderboard/terminal-bench/2.0 · gemini-3-flash-preview
Agent: Junie CLI
https://www.tbench.ai/leaderboard/terminal-bench/2.0
Original-source terms apply
Terminal Bench
51.7%
https://www.tbench.ai/leaderboard/terminal-bench/2.0 · gemini-3-flash-preview
Agent: Terminus 2
https://www.tbench.ai/leaderboard/terminal-bench/2.0
Original-source terms apply

Benchmark compilation: Epoch AI, Capabilities & benchmarking (CC BY attribution). Scores are converted to percentages and model names matched to the Vertical Tabs catalog. Individual evaluations are credited to their original sources. External results may carry separate licences and evaluation configurations; do not interpret scores from different harnesses as directly comparable.

Specifications

Model family
Gemini
Type
Language model
Predecessor
—
Successor
—
Released
December 17, 2025
Model ID
google/gemini-3-flash-preview
Openness
Closed
License
—
Input modalities
text, image, file, audio, video
Output modalities
text
Modality
text+image+file+audio+video->text
Tokenizer
Gemini
Max output tokens
66K
Knowledge cutoff
2025-01
Hugging Face
—
Source freshness
Metadata synced Oct 8, 2026, 10:17 PM UTC
Additional source evidence
Source model
google/gemini-3-flash-preview
Source family
gemini-flash
Last updated
2025-12-17
Knowledge cutoff
2025-01
Model context
1.0M
Model output limit
66K
Open weights
No
File attachments
Yes
Reasoning
Yes
Tool calling
Yes
Structured output
Yes
Temperature control
Yes

Published benchmarks

BenchmarkScoreEvaluation detailsEvidence
SWE-Bench Pro34.63 · resolve ratepublicSource
SWE-Atlas Codebase QnA8.2 · scoreMini-SWE-AgentSource
SWE-Atlas Refactoring10 · scoreMini-SWE-AgentSource
SWE-Atlas Test Writing30.3 · scoreMini-SWE-AgentSource
Complete source record (JSON)
{
  "id": "google/gemini-3-flash-preview",
  "name": "Gemini 3 Flash Preview",
  "description": "New Gemini flash lane bringing frontier-style multimodal reasoning to cheaper runs",
  "family": "gemini-flash",
  "last_updated": "2025-12-17",
  "knowledge": "2025-01",
  "attachment": true,
  "reasoning": true,
  "tool_call": true,
  "structured_output": true,
  "temperature": true,
  "modalities": {
    "input": [
      "text",
      "image",
      "video",
      "audio",
      "pdf"
    ],
    "output": [
      "text"
    ]
  },
  "limit": {
    "context": 1048576,
    "output": 65536
  },
  "benchmarks": [
    {
      "name": "SWE-Bench Pro",
      "score": 34.63,
      "source": "https://labs.scale.com/leaderboard/swe_bench_pro_public",
      "metric": "resolve rate",
      "dataset": "public"
    },
    {
      "name": "SWE-Atlas Codebase QnA",
      "score": 8.2,
      "source": "https://labs.scale.com/leaderboard/sweatlas-qna",
      "metric": "score",
      "harness": "Mini-SWE-Agent"
    },
    {
      "name": "SWE-Atlas Refactoring",
      "score": 10,
      "source": "https://labs.scale.com/leaderboard/sweatlas-refactoring",
      "metric": "score",
      "harness": "Mini-SWE-Agent"
    },
    {
      "name": "SWE-Atlas Test Writing",
      "score": 30.3,
      "source": "https://labs.scale.com/leaderboard/sweatlas-tw",
      "metric": "score",
      "harness": "Mini-SWE-Agent"
    }
  ],
  "release_date": "2025-12-17",
  "open_weights": false
}

Infrastructure

Provider availability

All providers
ProviderEndpointContextInput /M tokensOutput /M tokensCached input /M tokensCache write /M tokensUptime 1dData handling
Googlegoogle-vertex/global/flex1.0M$0.25$1.50$0.02$0.0499.93%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/global/flexBatch pricing1.0M$0.25$1.50$0.02$0.0499.93%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Google AI Studiogoogle-ai-studio/flex1.0M$0.25$1.50$0.02$0.04100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Google AI Studiogoogle-ai-studio/flexBatch pricing1.0M$0.25$1.50$0.02$0.04100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/global1.0M$0.50$3.00$0.05$0.08100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/globalBatch pricing1.0M$0.50$3.00$0.05$0.08100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Google AI Studiogoogle-ai-studio1.0M$0.50$3.00$0.05$0.08100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Google AI Studiogoogle-ai-studioBatch pricing1.0M$0.50$3.00$0.05$0.08100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/global/priority1.0M$0.90$5.40$0.09$0.15100.00%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Googlegoogle-vertex/global/priorityBatch pricing1.0M$0.90$5.40$0.09$0.15100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Google AI Studiogoogle-ai-studio/priority1.0M$0.90$5.40$0.09$0.15100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Google AI Studiogoogle-ai-studio/priorityBatch pricing1.0M$0.90$5.40$0.09$0.15100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Provider data:OpenRouter

Text provenance

Gemini app/web

Google SynthID text watermarking

Google DeepMind says text generated in the Gemini app and web experience is watermarked with SynthID by adjusting token probabilities.

Official explanation