ARC-AGIVerified
45.1%
ARC-AGI-2 semi-private
Models / Google
Gemini 3 Pro Deep Think
Gemini 3 · Released 2025-12-01
Google's extended-reasoning Gemini 3 Pro variant. ARC-AGI-2 semi-private 45.1%. For hardest puzzle-style reasoning, not default chat.
Reasoning-heavy Gemini route when standard Pro isn't enough on ARC-style tasks.
Composite35
Context1M
Input / 1M—
Output / 1M—
Knowledge cutoff2025-01-31
Statuslightweight
Not enough families covered yet.Add more benchmark rows for Gemini 3 Pro Deep Think to see the family profile.
Benchmark placements
Per-benchmark positions
Where Gemini 3 Pro Deep Think places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.
| Benchmark | Family | Rank | Score | Source | Date |
|---|---|---|---|---|---|
| CritPt | Reasoning | #8/ 12 | 26 | Artificial Analysis — CritPt | 2026-09-01 |
| ARC-AGI | Reasoning | #11/ 14 | 45 | ARC Prize leaderboard | 2026-05-15 |
Family context
Strongest rows in each family
The three highest-scoring models with pages in each capability family. Where Gemini 3 Pro Deep Think shows up, it's highlighted.
Knowledge
- 191
- 291
- 391
ReasoningGemini 3 Pro Deep Think · 45
- 195
- 295
- 395
Math
- 199
- 299
- 399
Coding
- 196
- 296
- 396
Agentic
- 1100
- 2100
- 3100
Long context
- 183
- 283
- 381
Tool use
- 189
- 287
- 387
Safety
- 192
- 292
- 391
Human preference
- 195
- 292
- 391
Newest receipts
2 recentLast 120 days
Benchmark rows added to the public ledger for Gemini 3 Pro Deep Think in the last 120 days. Older rows live in the full table below.
| Benchmark | Family | Score | Source | Days ago |
|---|---|---|---|---|
| CritPt | Reasoning | 25.7% | Artificial Analysis — CritPt | 5 |
| ARC-AGI | Reasoning | 45.1% | ARC Prize leaderboard | 114 |
Family coverage
Best score per family
A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.
Price vs. performance
Where Gemini 3 Pro Deep Think sits on the scatter
Hover any dot for name, score, and input price. Keyboard: tab through the top twelve, or use the ranking below.
AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMetaMiniMaxxAIZhipuXiaomiOtherMeituanCohereNVIDIA
Full benchmark ledger
Every catalog benchmark for Gemini 3 Pro Deep Think. Scores link to the original source; gaps mean no public row exists yet.
2 of 37 catalog benchmarks have a sourced row for Gemini 3 Pro Deep Think.
Reasoning
CritPtVerified
25.7%
CritPt composite (AA run)
Sources
Where the numbers come from
- ARC Prize leaderboard· checked 2026-06-09