Zhipu AI Open weights: zai-org/GLM-5.1Reasoning modelAuto-detected

GLM-5.1

Name: GLM-5.1 benchmark results
Creator: BenchLeader

Best configuration ranks #92 of 610 on the BenchLeader Index at 60.1 ±2.9. Last measured 8 Sept 2026. Released 7 Apr 2026.

Blended price: $2.15/M; $1.40 in · $4.40 out
Output speed: 60 tok/s
First answer: 66 s; first token 1.35 s
Context: 200k

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

Overall index60
Reasoning61
Coding57
Agents & tools57
Maths53
Knowledge57
Instruction following74
Human preference66
Long context63
Composite62

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
no reasoning	52.6	#244	38 tok/s	1.97 s	$0.0019	Agents & tools 65 · Composite 60 · Instruction following 53 · Knowledge 53 · Long context 53 · Maths 41
defaultbest	60.1	#92	60 tok/s	66 s	$0.0019	Agents & tools 57 · Coding 57 · Composite 62 · Human preference 66 · Instruction following 74 · Knowledge 57 · Long context 63 · Maths 53 · Reasoning 61

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	no reasoning	default	Source
GPQA Diamond	–	89.9%#39	Epoch AI Benchmarking Hub
SimpleBench	–	55.1%#39	SimpleBench
LMArena Hard Prompts	–	1489#42	LMArena
GPQA Diamond (AA)not in index	83.9%#150	86.8%#106	Artificial Analysis
Humanity's Last Exam (AA)not in index	27.9%#121	30.1%#103	Artificial Analysis
GPQA Diamond (Vals)not in index	–	84.5%#50	Vals AI

Coding

Benchmark	no reasoning	default	Source
SWE-bench Verified (Epoch)	–	74.2%#16	Epoch AI Benchmarking Hub
SciCode	–	43.8%#90	SciCode
WeirdML	–	57.1%#52	WeirdML
LMArena Coding	–	1514#37	LMArena
LMArena WebDev	–	1508#39	LMArena
SciCode (AA)not in index	–	44.8%#94	Artificial Analysis
LiveCodeBench	–	81.4%#60	Vals AI
SWE-bench (Vals)not in index	–	76.4%#35	Vals AI

Agents & tools

Benchmark	no reasoning	default	Source
Terminal-Bench Hard	35.6%#72	43.2%#39	Artificial Analysis
τ²-Bench Telecom (AA)not in index	97.1%#13	97.7%#9	Artificial Analysis
Terminal-Bench 2.1 (Vals)	–	56.9%#38	Vals AI
MCP Atlas	–	75.6%#15	Scale AI SEAL

Maths

Benchmark	no reasoning	default	Source
FrontierMath Tiers 1–3	24.9%#65	36.8%#50	Epoch AI Benchmarking Hub
OTIS Mock AIME	–	93.3%#44	Epoch AI Benchmarking Hub
ProofBench	–	22.2%#32	Vals AI
AIME (Vals)	–	91.9%#21	Vals AI
AIME 2026	–	95.8%#13	MathArena
HMMT February 2026	–	89.4%#14	MathArena
MathArena Apex	–	11.5%#20	MathArena

Knowledge

Benchmark	no reasoning	default	Source
SimpleQA Verified	–	34.0%#46	Epoch AI Benchmarking Hub
AA-Omniscience	-22.4#189	0.8#88	Artificial Analysis
MMLU-Pro	–	86.9%#36	Vals AI
LegalBench	–	84.4%#30	Vals AI
CorpFin	–	64.5%#37	Vals AI
TaxEval	–	71.2%#79	Vals AI

Instruction following

Benchmark	no reasoning	default	Source	Trend
IFBench	52.0%#161	76.3%#22	Artificial Analysis

Human preference

Benchmark	no reasoning	default	Source	Trend
LMArena Text	–	1466#44	LMArena

Long context

Benchmark	no reasoning	default	Source	Trend
AA-LCR	53.3%#241	73.7%#125	Artificial Analysis

Composite

Benchmark	no reasoning	default	Source	Trend
Epoch Capabilities Indexnot in index	–	149.7#46	Epoch AI Benchmarking Hub
AA Intelligence Index	24.2#125	26.4#106	Artificial Analysis

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
Friendli	99 tok/s	0.30 s	$1.40	$4.40	203k	–
Crusoe	78 tok/s	0.56 s	$1.20	$4.40	203k	fp8
Baidu Qianfan	60 tok/s	0.97 s	$0.965	$3.03	203k	fp8
AtlasCloud	58 tok/s	1.21 s	$1.26	$3.96	203k	fp8
NovitaAI	57 tok/s	2.06 s	$1.38	$4.40	205k	fp8
GMICloud	55 tok/s	1.51 s	$1.40	$4.40	203k	fp8
DeepInfra	53 tok/s	0.91 s	$1.05	$3.50	203k	fp4
Venice	51 tok/s	1.30 s	$1.54	$4.84	200k	fp8
Alibaba Cloud Int.	50 tok/s	1.52 s	$1.33	$4.18	203k	fp8
StreamLake	37 tok/s	1.35 s	$0.966	$3.04	200k	fp8
SiliconFlow	28 tok/s	1.66 s	$1.19	$3.74	205k	fp8
Nebius Token Factory	28 tok/s	0.96 s	$1.40	$4.40	203k	fp8
Chutes	24 tok/s	5.73 s	$0.980	$3.08	203k	fp8
Phala	21 tok/s	3.93 s	$1.21	$4.20	203k	–
Z.ai	20 tok/s	7.91 s	$1.40	$4.40	203k	fp8

Price history

Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.

input outputnow $0.965 in · $3.03 out

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 $0.260 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0019	$0.0015	1.2 min
Summarise a 30-page report	12,000 / 600	$0.019	$0.0092	1.3 min
Code edit	6,000 / 1,500	$0.015	$0.0099	1.5 min
Agentic coding session	60,000 / 4,000	$0.102	$0.050	2.2 min
Structured extraction	2,000 / 200	$0.0037	$0.0020	1.1 min