xAIReasoning modelAuto-detected

Grok 4.6

Name: Grok 4.6 benchmark results
Creator: BenchLeader

Best configuration ranks #42 of 610 on the BenchLeader Index at 63.8 ±6.5 (medium reasoning effort). Last measured 12 Aug 2026. Released 12 Aug 2026.

Blended price: $3.00/M; $2.00 in · $6.00 out
Output speed: 55 tok/s
First answer: 37 s; first token 1.31 s
Context: 500k

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

Overall index64
Reasoning62
Coding64
Knowledge77
Long context67
Composite83

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
low	59.6	#99	57 tok/s	7.71 s	$0.0026	Coding 55 · Composite 74 · Knowledge 76 · Long context 67 · Reasoning 50
mediumbest	63.8	#42	55 tok/s	37 s	$0.0026	Coding 64 · Composite 83 · Knowledge 77 · Long context 67 · Reasoning 62
high	58.7	#118	55 tok/s	41 s	$0.0029	Agents & tools 45 · Coding 66 · Knowledge 60 · Maths 67 · Reasoning 65
xhigh	62.3	#56	55 tok/s	48 s	$0.0026	Agents & tools 55 · Coding 60 · Composite 85 · Knowledge 68 · Long context 67 · Maths 61 · Reasoning 61
default	63.6	#44	55 tok/s	41 s	$0.0026	Agents & tools 69 · Coding 65 · Composite 73 · Knowledge 79 · Long context 67 · Maths 59 · Reasoning 71

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	low	medium	high	xhigh	default	Source
GPQA Diamond	–	–	94.0%#8	93.2%#16	–	Epoch AI Benchmarking Hub
SimpleBench	–	–	–	–	75.9%#8	SimpleBench
LiveBench Reasoningnot in index	–	–	–	–	90.5%#7	LiveBench
GPQA Diamond (AA)not in index	87.9%#92	93.5%#14	–	93.5%#14	95.0%#4	Artificial Analysis
Humanity's Last Exam (AA)not in index	27.6%#126	42.1%#44	–	44.1%#30	42.9%#34	Artificial Analysis
GPQA Diamond (Vals)not in index	–	–	94.7%#3	–	–	Vals AI
ARC-AGI-1	74.8%#93	87.5%#59	87.0%#63	87.0%#63	–	ARC Prize
ARC-AGI-2	27.6%#91	61.3%#54	65.1%#47	67.1%#43	–	ARC Prize
ARC-AGI-3	–	–	–	2.1%#17	–	ARC Prize

Coding

Benchmark	low	medium	high	xhigh	default	Source
SciCode	48.4%#71	54.6%#28	53.6%#37	51.6%#49	–	SciCode
WeirdML	–	–	67.3%#34	–	–	WeirdML
FrontierCode	–	–	–	–	48.0%#5	Cognition
LMArena WebDev	–	–	1624#11	–	–	LMArena
LiveBench Codingnot in index	–	–	–	–	76.8%#31	LiveBench
SciCode (AA)not in index	49.4%#75	55.9%#26	–	53.0%#50	56.5%#20	Artificial Analysis
LiveCodeBench	–	–	88.2%#7	–	–	Vals AI
SWE-bench (Vals)not in index	–	–	95.6%#4	–	–	Vals AI

Agents & tools

Benchmark	low	medium	high	xhigh	default	Source
Terminal-Bench	–	–	20.3%#55	–	–	Terminal-Bench
APEX-Agents	–	–	–	–	41.2%#8	Mercor
LMArena Agent	–	–	–	3.2#16	–	LMArena
LiveBench Agentic Codingnot in index	–	–	–	–	57.0%#18	LiveBench
Terminal-Bench 2.1 (Vals)	–	–	78.3%#10	–	–	Vals AI

Maths

Benchmark	low	medium	high	xhigh	default	Source
FrontierMath Tiers 1–3	–	–	–	66.0%#21	–	Epoch AI Benchmarking Hub
FrontierMath Tier 4	–	–	–	31.7%#25	–	Epoch AI Benchmarking Hub
OTIS Mock AIME	–	–	97.8%#17	99.2%#10	–	Epoch AI Benchmarking Hub
ProofBench	–	–	–	–	51.0%#16	Vals AI
LiveBench Mathematicsnot in index	–	–	–	–	92.6%#16	LiveBench

Knowledge

Benchmark	low	medium	high	xhigh	default	Source
SimpleQA Verified	–	–	49.3%#23	48.9%#24	–	Epoch AI Benchmarking Hub
LiveBench Data Analysisnot in index	–	–	–	–	73.9%#30	LiveBench
AA-Omniscience	25.9#29	28#24	–	29.3#19	30.5#17	Artificial Analysis
MMLU-Pro	–	–	89.4%#11	–	–	Vals AI
LegalBench	–	–	86.3%#12	–	–	Vals AI
CorpFin	–	–	66.2%#20	–	–	Vals AI
TaxEval	–	–	71.1%#82	–	–	Vals AI

Instruction following

Benchmark	low	medium	high	xhigh	default	Source	Trend
LiveBench Languagenot in index	–	–	–	–	83.7%#13	LiveBench

Long context

Benchmark	low	medium	high	xhigh	default	Source	Trend
AA-LCR	80.7%#41	81.0%#37	–	81.0%#37	80.3%#45	Artificial Analysis

Composite

Benchmark	low	medium	high	xhigh	default	Source
Epoch Capabilities Indexnot in index	–	–	–	–	156.3#16	Epoch AI Benchmarking Hub
LiveBench	–	–	–	–	78.0%#11	LiveBench
AA Intelligence Index	35.4#51	43.0#25	–	44.3#22	44.4#21	Artificial Analysis
Vals Indexnot in index	–	–	59.2#14	–	–	Vals AI

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
SpaceXAI (ZDR)	52 tok/s	1.36 s	$2.00	$6.00	500k	–
SpaceXAI	52 tok/s	0.96 s	$2.00	$6.00	500k	–
SpaceXAI Priority (ZDR)	52 tok/s	1.31 s	$4.00	$12.00	500k	–
SpaceXAI Priority	50 tok/s	1.19 s	$4.00	$12.00	500k	–
Amazon Bedrock (US)	17 tok/s	24 s	$2.20	$6.60	500k	–

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 $0.550 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0026	$0.0022	42.2 s
Summarise a 30-page report	12,000 / 600	$0.028	$0.015	47.6 s
Code edit	6,000 / 1,500	$0.021	$0.015	1.1 min
Agentic coding session	60,000 / 4,000	$0.144	$0.079	1.8 min
Structured extraction	2,000 / 200	$0.0052	$0.0030	40.3 s