Gemini 3.1 Flash Lite vs Gemini 3.5 Flash Lite
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Gemini 3.1 Flash Lite17.2
Gemini 3.5 Flash Lite18.8
Terminal-Bench 4.0 Terminal-Bench 4.0 · source setting 2: Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
Gemini 3.1 Flash Lite0.5
Gemini 3.5 Flash Lite1
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
Gemini 3.1 Flash Lite6.8
Gemini 3.5 Flash Lite25
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
Gemini 3.1 Flash Lite—
Gemini 3.5 Flash Lite1,436
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos
Gemini 3.1 Flash Lite75.5
Gemini 3.5 Flash Lite79
| Benchmark | Gemini 3.1 Flash Lite | Gemini 3.5 Flash Lite |
|---|---|---|
| Humanity's Last Exam | 17.2 | 18.8 |
| Terminal-Bench 4.0 | 0.5 | 1 |
| AutomationBench-AA | 6.8 | 25 |
| Arena Text · creative writing | — | 1,436 |
| MMMU-Pro (vision) | 75.5 | 79 |
Specs
| Spec | Gemini 3.1 Flash Lite | Gemini 3.5 Flash Lite |
|---|---|---|
| Input, per 1M | $0.25 | $0.30 |
| Output, per 1M | $1.50 | $2.50 |
| Cost per task? | $0.06 | $0.19 |
| Speed, tok/s | 366 | 454 |
| Context | 1M | 1M |
| Weights | Closed | Closed |
| Released | May 2026 | Jul 2026 |
| See images | ✓ | ✓ |
| Read PDFs | ✓ | ✓ |
| Hear audio | — | ✓ |
| Watch video | ✓ | ✓ |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ✓ | ✓ |
| Run code | ✓ | ✓ |
| Use a computer | — | ? |
What breaks when you switch
Gemini 3.1 Flash Lite → Gemini 3.5 Flash Lite
Nothing known.
Gemini 3.5 Flash Lite → Gemini 3.1 Flash Lite
Nothing known.