GoogleReasoning modelAuto-detected

Gemini 3.1 Pro

Name: Gemini 3.1 Pro benchmark results
Creator: BenchLeader

Best configuration ranks #37 of 610 on the BenchLeader Index at 63.9 ±3.9. Last measured 8 Sept 2026. Released 19 Feb 2026.

Blended price: $4.50/M; $2.00 in · $12.00 out
Output speed: 110 tok/s
First answer: 30 s; first token 5.50 s
Context: 1.0M

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Overall index64
Reasoning70
Coding60
Agents & tools64
Maths61
Knowledge66
Instruction following70
Human preference60
Multimodal66
Long context68
Composite67

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.

Effort	Index	Rank	Speed	First answer	Chat reply cost	Categories
high	60.7	#76	110 tok/s	30 s	$0.0044	Agents & tools 58 · Coding 64 · Composite 57 · Knowledge 66 · Maths 66 · Reasoning 65
thinking	–	–	110 tok/s	30 s	$0.0044	Coding 60
defaultbest	63.9	#37	110 tok/s	30 s	$0.0044	Agents & tools 64 · Coding 60 · Composite 67 · Human preference 60 · Instruction following 70 · Knowledge 66 · Long context 68 · Maths 61 · Multimodal 66 · Reasoning 70

Benchmark results

One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	high	thinking	default	Source
GPQA Diamond	94.4%#5	–	94.1%#7	Epoch AI Benchmarking Hub
Humanity's Last Exam	–	–	46.4%#2	Scale AI / CAIS
SimpleBench	–	–	79.6%#3	SimpleBench
LMArena Hard Prompts	–	–	1507#16	LMArena
LiveBench Reasoningnot in index	84.0%#31	–	–	LiveBench
GPQA Diamond (AA)not in index	–	–	94.1%#7	Artificial Analysis
Humanity's Last Exam (AA)not in index	–	–	47.0%#23	Artificial Analysis
GPQA Diamond (Vals)not in index	95.5%#1	–	–	Vals AI
EnigmaEval	36.8%#3	–	–	Scale AI SEAL
ARC-AGI-1	–	–	98.0%#5	ARC Prize
ARC-AGI-2	–	–	77.1%#28	ARC Prize
ARC-AGI-3	–	–	0.4%#24	ARC Prize

Coding

Benchmark	high	thinking	default	Source
SciCode	–	–	58.9%#4	SciCode
WeirdML	–	–	72.1%#28	WeirdML
GSO-Bench	–	–	22.6%#12	GSO-Bench
LMArena Coding	–	–	1521#25	LMArena
LMArena WebDev	–	–	1447#54	LMArena
LiveBench Codingnot in index	76.5%#32	–	–	LiveBench
SciCode (AA)not in index	–	–	58.7%#10	Artificial Analysis
LiveCodeBench	88.5%#6	–	–	Vals AI
SWE-bench (Vals)not in index	78.8%#28	–	–	Vals AI
SWE-Bench Pro	–	46.1%#5	–	Scale AI SEAL

Agents & tools

Benchmark	high	thinking	default	Source
Terminal-Bench	–	–	80.2%#3	Terminal-Bench
APEX-Agents	–	–	33.5%#22	Mercor
LMArena Agent	–	–	-5.3#33	LMArena
LiveBench Agentic Codingnot in index	44.1%#42	–	–	LiveBench
Terminal-Bench Hard	–	–	53.8%#16	Artificial Analysis
τ²-Bench Telecom (AA)not in index	–	–	95.6%#20	Artificial Analysis
Terminal-Bench 2.1 (Vals)	70.8%#20	–	–	Vals AI
MCP Atlas	78.2%#12	–	–	Scale AI SEAL
HiL-Bench	–	–	35.3%#9	Scale AI SEAL

Maths

Benchmark	high	thinking	default	Source
FrontierMath Tiers 1–3	–	–	59.6%#27	Epoch AI Benchmarking Hub
FrontierMath Tier 4	–	–	26.8%#31	Epoch AI Benchmarking Hub
OTIS Mock AIME	95.6%#29	–	95.6%#28	Epoch AI Benchmarking Hub
ProofBench	–	–	26.0%#30	Vals AI
LiveBench Mathematicsnot in index	91.0%#20	–	–	LiveBench
AIME (Vals)	98.1%#1	–	–	Vals AI
AIME 2026	–	–	98.3%#4	MathArena
HMMT February 2026	–	–	94.7%#8	MathArena
MathArena Apex	–	–	60.9%#5	MathArena

Knowledge

Benchmark	high	thinking	default	Source
SimpleQA Verified	73.5%#2	–	–	Epoch AI Benchmarking Hub
LiveBench Data Analysisnot in index	78.5%#15	–	–	LiveBench
AA-Omniscience	–	–	31.9#15	Artificial Analysis
MMLU-Pro	91.0%#4	–	–	Vals AI
LegalBench	87.4%#3	–	–	Vals AI
CorpFin	64.5%#36	–	–	Vals AI
TaxEval	72.9%#57	–	–	Vals AI
MedQA	96.4%#3	–	–	Vals AI
PRBench Finance	–	–	41.9%#21	Scale AI SEAL
PRBench Legal	–	–	44.0%#17	Scale AI SEAL
MultiNRC	–	–	64.7%#3	Scale AI SEAL

Instruction following

Benchmark	high	thinking	default	Source
LiveBench Languagenot in index	85.4%#10	–	–	LiveBench
IFBench	–	–	77.1%#18	Artificial Analysis
MultiChallenge	–	–	71.4%#3	Scale AI SEAL
TutorBench	–	–	53.0%#12	Scale AI SEAL

Human preference

Benchmark	high	thinking	default	Source	Trend
LMArena Text	–	–	1487#15	LMArena
EQ-Bench 4	–	–	1142#20	EQ-Bench

Multimodal

Benchmark	high	thinking	default	Source
LMArena Vision	–	–	1295#17	LMArena
MMMU-Pro	–	–	82.4%#19	Artificial Analysis
MMMU-Pro (official)not in index	80.5%#2	–	–	MMMU

Long context

Benchmark	high	thinking	default	Source	Trend
AA-LCR	–	–	82.0%#25	Artificial Analysis

Composite

Benchmark	high	thinking	default	Source
Epoch Capabilities Indexnot in index	–	–	155#25	Epoch AI Benchmarking Hub
LiveBench	77.0%#17	–	–	LiveBench
AA Intelligence Index	–	–	30.4#80	Artificial Analysis
Vals Indexnot in index	41.9#34	–	–	Vals AI

Where it wins

Benchmarks where this configuration ranks in the top five of every configuration measured.

Where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

Provider	Speed	First token	Input $/M	Output $/M	Context	Quantisation
Google AI Studio Flex	115 tok/s	7.05 s	$1.00	$6.00	1.0M	–
Google Vertex	102 tok/s	2.88 s	$2.00	$12.00	1.0M	–
Google AI Studio	90 tok/s	5.50 s	$2.00	$12.00	1.0M	–

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.200 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0044	$0.0039	33.2 s
Summarise a 30-page report	12,000 / 600	$0.031	$0.015	35.9 s
Code edit	6,000 / 1,500	$0.030	$0.022	44.1 s
Agentic coding session	60,000 / 4,000	$0.168	$0.087	1.1 min
Structured extraction	2,000 / 200	$0.0064	$0.0037	32.3 s