Gemini 3.5 Flash
Gemini 3.5 FlashGoogleProprietaryReasoningVerifiedReleased May 19, 2026Input: text, image, video, file, audio
Overall score
61.9
Rank 50 of 222 · 53.7–70.1
Context window
1M
1,000,000 tokens
Price per 1M tokens
$1.50 in · $9.00 out
Source: BenchLM
Blended price
$3.38
3:1 input:output tokens
Output speed
—
Not measured yet
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Multimodal#787.8
- 2Instruction following#3787.4
- 3Knowledge#3272.4
- 4Agentic & tool use#4771.8
- 5Reasoning#1670.9
- 6Math67.1
- 7Coding#4451.7
Benchmark results (26)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Agentic & tool use
| Terminal-Bench 2.1 | 76.2 | #34 of 68 |
| Gert Labs | 61.85 | #10 of 52 |
| MCP Atlas | 83.6 | #7 of 42 |
| MazeBench | 0 | #20 of 42 |
| OSWorld-Verified | 78.4 | #13 of 35 |
| Toolathlon | 56.5 | #4 of 22 |
| ResearchClawBench | 18.0 | #8 of 19 |
Multimodal
| Design Arena Website | 1268 | #30 of 91 |
| MMMU-Pro | 83.6 | #2 of 48 |
| CharXiv | 84.2 | #16 of 42 |
| ParseBench | 74.02 | #5 of 37 |
| ExtractBench (LlamaIndex)(lower is better) | 79.84 | #6 of 24 |
| Blueprint-Bench 2 | 33.6 | #2 of 2 |
Reasoning
| ARC-AGI-2 | 72.1 | #20 of 41 |
| ARC-AGI-1 | 92.50 | #18 of 36 |
| MRCRv2 | 77.3 | #4 of 12 |
| MRCR 1M | 26.6 | #4 of 4 |
Coding
| SWE-bench Pro | 55.1 | #48 of 69 |
| SciCode | 53.1 | #3 of 23 |
Knowledge
| GPQA | 92.2 | #16 of 117 |
| HLE | 40.2 | #31 of 69 |
Math
| FrontierMath v2 (Tiers 1-3) | 62.807 | #32 of 64 |
| FrontierMath v2 (Tier 4) | 26.829 | #30 of 51 |
Instruction following
| IFBench | 76.3 | #20 of 42 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| List price (BenchLM) | $1.50 | $9.00 | $0.15 |
| OpenRouter google/gemini-3.5-flash | $1.50 | $9.00 | $0.15 |
Google's official Gemini Developer API pricing page lists Gemini 3.5 Flash at $1.50 input / $0.15 cached input / $9.00 output per million tokens on the standard paid tier.
Arena ratings
Human-preference ratings from LMArena. Best-rated variant shown.
| Text(high) | 1482 | #18 · 48.5K votes |
| WebDev (coding)(High) | 1499 | #62 · 9.4K votes |
| Vision(High) | 1308 | #13 · 13.9K votes |
| Documents(medium) | 1463 | #20 · 6.5K votes |
Speed
Measured by us through OpenRouter (median of trials).
Not measured yet.
Related models
Same family
- Gemini 3 Flash54.3
- Gemini 3.1 Pro64.7
Similar score
- Mistral Large 461.6 · $1.03
- Ornith-1.5-397B62.3 · —
- Claude Opus 4.662.3 · $10.00
- Hy4 preview61.4 · $1.13
- Qwen3.7 Max62.6 · $2.21