Gemini 3.1 Pro Preview vs Gemini 4 Argon

$1.30 / task
Wins 0 of 5 benchmarks
$1.99 / task
Wins 4 of 5 benchmarks
Different efforts: thinking · high · Compare all at high
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Gemini 3.1 Pro Preview47
Gemini 4 Argon57.1
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 Pro Preview4
Gemini 4 Argon57.1
AutomationBench-AA AutomationBench-AA: Agents orchestrate SaaS workflows through APIs; objectives receive credit only when the task has no guardrail violation
Gemini 3.1 Pro Preview35.4
Gemini 4 Argon77.5
Arena Text · creative writing Arena Text · creative writing: Human pairwise preferences in the stated text vote pool
Gemini 3.1 Pro Preview1,481
Gemini 4 Argon1,516
MMMU-Pro (vision) MMMU-Pro (vision): College-level questions that require understanding images: charts, diagrams, photos
Gemini 3.1 Pro Preview82.4
Gemini 4 Argon—

Specs

SpecGemini 3.1 Pro PreviewGemini 4 Argon
Input, per 1M$2!$2
Output, per 1M$12!$10
Cost per task?$1.30$1.99
Speed, tok/s11584
Context1M1M
WeightsClosedClosed
ReleasedFeb 2026Oct 2026
See images✓✓
Read PDFs✓✓
Hear audio✓✓
Watch video✓✓
Call tools✓✓
Force a tool callPartialPartial
Return strict JSON✓✓
Search the web✓✓
Run code✓✓
Use a computer??

What breaks when you switch

Gemini 3.1 Pro Preview → Gemini 4 Argon
Synchronous output stops at 65,536›

With background false, output limits above 65,536 tokens return HTTP 400; stay at or below that limit.

Sampling parameters are ignored›

temperature, top_p and top_k are ignored; remove them from requests.

No explicit caching›

Explicit caching is unavailable; omit explicit cache requests.

Gemini 4 Argon → Gemini 3.1 Pro Preview
Cannot use both thinking controls›

Setting thinking_level and legacy thinking_budget together returns HTTP 400; use only one.