Gemini 3 Pro: benchmark scores, price and ranking

Gemini 3 Pro from Google DeepMind ranks #180 of 743 LLMs on the LLMs Tiger aggregate benchmark score (50.6), based on 7 benchmark results captured up to 2026-10-13.

Organisation: Google DeepMind
Type: Frontier model
Open weights: No
Released: 2025-11-18 (source: curated)
Cheapest API price: $2.00 in / $12.00 out per million tokens (via replicate, observed 2026-10-13T20:47Z)
Aggregate rank: #180 of 743 (score 50.6)

Relative to other models, Gemini 3 Pro is strongest on Epoch ECI (#4 of 204) and weakest on LMArena Document Elo (#11 of 18).

Across the 7 benchmarks where at least ten models are scored, Gemini 3 Pro is in the top 10% on 1 and in the top half on 6.

Its cheapest listed input price, $2.00 per million tokens, is lower than that of 18% of the 427 models with a listed price.

Gemini 3 Pro benchmark results

Gemini 3 Pro benchmark scores, ranked against all models measured on each benchmark
Benchmark	Score	Rank	Source	Captured
Epoch ECI	153	#4 of 204	Epoch AI Benchmarking Hub	2026-10-13
SWE-bench Verified	69.6%	#21 of 73	SWE-bench Verified leaderboard	2026-10-13
LMArena Vision Elo	1,289	#7 of 22	LMArena	2026-10-13
LMArena Text Elo	1,485	#10 of 28	LMArena	2026-10-13
SWE-bench Multilingual	68.7%	#5 of 11	SWE-bench Multilingual leaderboard	2026-10-13
ARC-AGI-2	31.1%	#43 of 87	ARC Prize official leaderboard	2026-10-13
LMArena Document Elo	1,434	#11 of 18	LMArena	2026-10-13

About Gemini 3 Pro

Gemini 3 Pro is a frontier model from Google DeepMind. It was released on 2025-11-18, according to the developer's announcement. The cheapest API listing we track charges $2.00 per million input tokens and $12.00 per million output tokens, through replicate. We hold 7 benchmark results for it across 3 areas, and 1 is in the top tenth of its board.

Where Gemini 3 Pro stands

On our aggregate score, Gemini 3 Pro is #180 of 743 ranked models, in the top quarter. Its score is 50.6; the scale is centred on 50, so it sits 0.6 points above the typical model. That beats 76% of the ranked field. Among the 41 ranked Google DeepMind models it is #10. The next one up is Gemini 2.5 Pro Preview (Mar 2025) at 50.9. The next one down is gemini-2.5-flash-preview-05-20 at 50.5.

Gemini 3 Pro, benchmark by benchmark

On Epoch ECI, Gemini 3 Pro scores 153 and ranks #4 of 204. Only 3 models score higher, led by Qwen 3.8 Max at 156, which is 3 points ahead. It covers Epoch Capabilities Index: Epoch's own aggregate of ~40 benchmarks. The board median is 127, so Gemini 3 Pro is 26 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On SWE-bench Verified, Gemini 3 Pro reaches 69.6%. It ranks #21 of 73, ahead of 71% of the field, and 13.9 points behind Claude Opus 4.7 (max) at 83.5%. The results either side are DeepSeek-V3.2 (70.0%) and GPT-5.1 (high) (68.0%). It covers real GitHub issues resolved. The board median is 56.4%, so Gemini 3 Pro is 13.2 points above it. Source: SWE-bench Verified leaderboard, captured 2026-10-13.

Gemini 3 Pro manages 1,289 on LMArena Vision Elo, good for #7 of 22. Gemini 3.7 Flash (high) leads at 1,296, 7 points further up, and 68% of models sit below. The results either side are Muse Spark 1.3 (max) (1,290) and Claude Fable 5.1 (max) (1,288). It covers crowd-voted image understanding. The board median is 1,279, so Gemini 3 Pro is 10 points above it. Source: LMArena, captured 2026-10-13.

LMArena Text Elo puts Gemini 3 Pro at 1,485, #10 of 28. That beats 64% of the models tested, but 40 points short of gemini-4-argon-high (1,525). The results either side are Kimi K3 (max) (1,488) and Gemini 3.6 Flash (High) (1,483). It covers head-to-head chat preference (crowd votes). The board median is 1,478, so Gemini 3 Pro is 8 points above it. Source: LMArena, captured 2026-10-13.

On SWE-bench Multilingual, Gemini 3 Pro reaches 68.7%. It ranks #5 of 11, ahead of 55% of the field, and 4.0 points behind Gemini 3 Flash at 72.7%. The results either side are GLM-5 (69.7%) and MiniMax-M2.5 (68.3%). It covers real GitHub issues resolved. The board median is 68.3%, so Gemini 3 Pro is 0.4 points above it. Source: SWE-bench Multilingual leaderboard, captured 2026-10-13.

Gemini 3 Pro manages 31.1% on ARC-AGI-2, good for #43 of 87. GPT-6 Astra (max) leads at 95.0%, 63.9 points further up, and 51% of models sit below. The results either side are Gemini 3 Flash Preview (High) (33.6%) and Opus 4.5 (Thinking, 16K) (22.8%). It covers abstract reasoning, ARC Prize public eval set. The board median is 22.8%, so Gemini 3 Pro is 8.3 points above it. Source: ARC Prize official leaderboard, captured 2026-10-13.

Gemini 3 Pro lands below the middle on LMArena Document Elo: 1,434, #11 of 18, with 56% of the field scoring higher. Claude Fable 5.1 (max) leads at 1,513; the closest result above is MiniMax-M3 at 1,435. It covers crowd-voted questions about documents. The board median is 1,440, so Gemini 3 Pro is 6 points below it. Source: LMArena, captured 2026-10-13.

Strengths and weaknesses

General knowledge and chat: solid, ahead of 65% of the field on average across 5 benchmarks. Its best board here is Epoch ECI (#4 of 204) and its weakest is LMArena Document Elo (#11 of 18).

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so Gemini 3 Pro costs more than the typical listing. Models that score within 1.5 points or better and cost less: Qwen2.5-Coder-7B-Instruct ($0.010, 49.5), DeepSeek V4 Flash 0731 (max) ($0.016, 53.1), GLM-5.3-Flash (max) ($0.026, 49.6). Similar scores cost more on gpt-4o-2024-11-20 ($2.50) and gpt-4o-2024-08-06 ($2.50).

Release history

The Google DeepMind model released before it was gemini-2.5-pro_16K on 2025-06-17, which scores 50.4. The next one was Gemini 3 Flash on 2025-12-17, at 51.5.

Reading these numbers

The 7 scores on this page come from 5 sources: LMArena (3), ARC Prize official leaderboard (1), Epoch AI Benchmarking Hub (1), SWE-bench Multilingual leaderboard (1). 2 of them come from an official leaderboard and 5 from datasets or third-party runs. With 7 results the aggregate rank can move a lot when one new score arrives.

Models ranked nearby

#177 MiniMax-Text-01 (50.6)
#178 Claude Sonnet 4.5 (59k thinking) (50.6)
#179 chatgpt-4o-01-29-2025 (50.6)
#181 NVIDIA-Nemotron-3-Super-120B-A12B-FP8 (50.5)
#182 Claude 3.7 Sonnet w/ Review Heavy (50.5)
#183 GPT-5-codex (high) (50.5)

More from Google DeepMind

Compare it on the live leaderboard → · How the aggregate score works