AnthropicReasoning modelCompare with another model

Claude Opus 4.7

Reasoning effort

Claude Opus 4.7 is an Anthropic proprietary reasoning model, released 16 Apr 2026. Its best configuration ranks #63 of 775 on the BenchLeader Index at 63.4 ±2.6, in the upper half. The ± is the point: 143 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in human preference (68) and lowest in coding (62). At $10.00 per million tokens blended it is among the most expensive ranked models. Output speed of 60 tokens per second puts it slower than most, with a first token in 1.2 s. It has been measured at 7 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 19 Oct 2026.

Blended price: $10.00/M; $5.00 in · $25.00 out
Output speed: 60 tok/s; OpenRouter traffic, 7-day median; not yet measured by Artificial Analysis
First answer: 1.23 s
Context: 1M
Full answer: –
Cost per run: –
Released: 16 Apr 2026

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

Overall index63
Reasoning65
Coding62
Agents & tools66
Maths67
Knowledge62
Human preference68
Multimodal67

Versions

Anthropic has shipped 9 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

Model	Released	Index	Rank
Claude Opus 5.5max	22 Sept 2026	72.7	#1
Claude Opus 5max	24 Jul 2026	68.9	#15
Claude Opus 4.8max	28 May 2026	63.4	#64
Claude Opus 4.7this page	16 Apr 2026	63.4	#63
Claude Opus 4.6high	5 Feb 2026	62.9	#76
Claude Opus 4.5thinking	24 Nov 2025	60.0	#119
Claude Opus 4.1thinking	5 Aug 2025	56.5	#199
Claude Opus 4thinking	22 May 2025	55.1	#230
Claude 3 Opus	4 Mar 2024	39.2	#681

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. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

Effort	Index	Rank	Speed	First answer	Chat reply cost	Categories
no reasoning	58.1	#164	50 tok/s	1.56 s	$0.0095	Agents & tools 81 · Composite 64 · Instruction following 46 · Knowledge 68 · Long context 63 · Multimodal 59 · Reasoning 49
low	–	–	60 tok/s	1.23 s	$0.0095
high	61.2	#99	60 tok/s	1.23 s	$0.0095	Coding 64 · Human preference 70 · Maths 69 · Multimodal 66 · Reasoning 55
xhigh	57.4	#178	60 tok/s	1.23 s	$0.0095	Knowledge 58 · Maths 60 · Reasoning 65
max	62.4	#86	54 tok/s	31 s	$0.0095	Agents & tools 60 · Coding 65 · Composite 75 · Instruction following 59 · Knowledge 74 · Long context 64 · Maths 58 · Multimodal 62 · Reasoning 62
thinking	–	–	60 tok/s	1.23 s	$0.0095	Reasoning 73
not statedbest	63.4	#63	60 tok/s	1.23 s	$0.0095	Agents & tools 66 · Coding 62 · Human preference 68 · Knowledge 62 · Maths 67 · Multimodal 67 · Reasoning 65

How its index has moved

8 Sept 2026 to 19 Oct 2026

not stated
max
high
no reasoning

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. 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	low	high	xhigh	max	thinking	not stated	Source
GPQA Diamond	–	–	–	90.2%#44	86.4%#77	–	–	Epoch AI Benchmarking Hub
Humanity's Last Exameffort not stated	–	–	–	–	–	–	36.2%#10	Scale AI / CAIS
SimpleBencheffort not stated	–	–	–	–	–	–	61.7%#23	SimpleBench
Arena Hard Prompts	–	–	1525#7	–	–	–	1519#9	Arena (LMArena)
LiveBench 1.0 Reasoningnot in index	–	–	–	87.2%#28	–	–	–	LiveBench
GPQA Diamond (AA)not in index	88.5%#88	–	–	–	91.4%#47	–	–	Artificial Analysis
Humanity's Last Exam (AA)not in index	33.3%#134	–	–	–	42.3%#69	–	–	Artificial Analysis
GPQA Diamond (Vals)not in indexeffort not stated	–	–	–	–	–	–	90.2%#25	Vals AI
Kagi LLM Benchmark	–	–	–	–	–	80.7%#7	73.3%#18	Kagi LLM Benchmark
ARC-AGI-3	–	–	0.2%#54	–	–	–	–	ARC Prize
CritPt	5.1%#151	–	–	–	12.0%#107	–	–	Artificial Analysis
Chess Puzzlesnot in index	–	26.0%#75	–	36.0%#48	41.0%#39	–	–	Epoch AI Benchmarking Hub
Mystery Game Puzzlesnot in index	–	–	–	–	28.0%#38	–	13.0%#101	Epoch AI Benchmarking Hub
LMCAnot in index	–	–	–	–	52.2%#56	–	–	Epoch AI Benchmarking Hub
DTBenchnot in index	–	–	–	–	94.7%#55	–	–	Epoch AI Benchmarking Hub
ForecastBenchnot in indexeffort not stated	–	–	–	–	–	–	63.1%#48	ForecastBench

Coding

Benchmark	high	xhigh	max	not stated	Source
SWE-bench Verified (Epoch)	–	–	83.5%#1	–	Epoch AI Benchmarking Hub
SciCode	–	–	54.5%#60	–	SciCode
WeirdML	76.4%#25	–	75.5%#28	76.4%#26	WeirdML
FrontierCodeeffort not stated	–	–	–	38.5%#29	Cognition
GSO-Bench	44.1%#6	–	–	44.1%#6	GSO-Bench
Arena Coding	1551#3	–	–	1546#8	Arena (LMArena)
Arena WebDev	1558#37	–	–	1556#39	Arena (LMArena)
LiveBench 1.0 Codingnot in index	–	82.1%#11	–	–	LiveBench
LiveCodeBencheffort not stated	–	–	–	85.1%#34	Vals AI
SWE-bench (Vals)not in indexeffort not stated	–	–	–	82.0%#21	Vals AI
Code Migrationnot in indexeffort not stated	–	–	–	43.9%#25	Vals AI
ProgramBenchnot in indexeffort not stated	–	–	–	0.0%#32	Vals AI
Vibe Code Bench v1.1not in indexeffort not stated	–	–	–	71.0%#39	Vals AI
SWE Atlas: Codebase QnAnot in indexeffort not stated	–	–	–	40.3%#11	Scale AI SEAL
SWE Atlas: Refactoringnot in indexeffort not stated	–	–	–	48.6%#4	Scale AI SEAL
SWE Atlas: Test Writingnot in indexeffort not stated	–	–	–	38.5%#13	Scale AI SEAL
MirrorCodenot in index	31.1%#5	–	–	–	Epoch AI Benchmarking Hub
ALE-Benchnot in indexeffort not stated	–	–	–	1323.0#30	Epoch AI Benchmarking Hub
Terminal-Bench 2.1 (AA)not in index	–	–	83.2%#36	–	Artificial Analysis

Agents & tools

Benchmark	no reasoning	xhigh	max	not stated	Source
Terminal-Bencheffort not stated	–	–	–	80.2%#3	Terminal-Bench
OSWorld-Verified 2.0	–	–	18.2%#8	–	OSWorld
APEX-Agents	–	–	49.2%#28	–	Mercor
LiveBench Agentic Codingnot in index	–	50.7%#42	–	–	LiveBench
Terminal-Bench Hard	54.5%#15	–	51.5%#21	–	Artificial Analysis
τ²-Bench Telecom (AA)not in index	74.0%#143	–	88.6%#71	–	Artificial Analysis
Terminal-Bench 2.1 (Vals)effort not stated	–	–	–	68.5%#31	Vals AI
MCP Atlas	–	–	79.1%#13	–	Scale AI SEAL
HiL-Bencheffort not stated	–	–	–	41.7%#15	Scale AI SEAL
τ³-bench Bankingnot in index	–	–	40.2%#7	–	τ²-bench
GDPval-AA v2.1not in index	–	–	42.6%#88	–	Artificial Analysis
τ³-Banking (AA)	–	–	34.6%#50	–	Artificial Analysis
ITBench SRE (AA)not in index	–	–	46.7%#11	–	Artificial Analysis
Analyst Agent (AA)not in index	–	–	43.8%#20	–	Artificial Analysis
Finance Agent v2not in indexeffort not stated	–	–	–	51.5%#39	Vals AI
Legal Research Benchnot in indexeffort not stated	–	–	–	38.5%#33	Vals AI
Tax Agent Benchnot in indexeffort not stated	–	–	–	23.7%#39	Vals AI
Terminal-Bench 2.0 (Vals)not in indexeffort not stated	–	–	–	68.5%#3	Vals AI
Arena Searchnot in indexeffort not stated	–	–	–	1210#12	Arena (LMArena)
PostTrainBenchnot in index	–	28.6%#8	–	–	Epoch AI Benchmarking Hub
ExploitBenchnot in indexeffort not stated	–	–	–	26.5%#3	Epoch AI Benchmarking Hub
GBAEvalnot in indexeffort not stated	–	–	–	43.8%#12	Epoch AI Benchmarking Hub
Vending-Bench 2not in indexeffort not stated	–	–	–	10936.8#6	Andon Labs

Maths

Benchmark	high	xhigh	max	not stated	Source
FrontierMath Tiers 1–3	–	–	70.2%#27	–	Epoch AI Benchmarking Hub
FrontierMath Tier 4	–	–	31.7%#35	–	Epoch AI Benchmarking Hub
OTIS Mock AIME	–	97.8%#29	86.7%#85	–	Epoch AI Benchmarking Hub
ProofBench	–	–	54.0%#27	–	Vals AI
LiveBench Mathematicsnot in index	–	92.8%#26	–	–	LiveBench
AIME (Vals)effort not stated	–	–	–	96.3%#7	Vals AI
AIME 2026possible contamination	–	95.8%#13	–	–	MathArena
HMMT February 2026possible contamination	–	93.9%#10	–	–	MathArena
MathArena Apexpossible contamination	–	40.6%#7	–	–	MathArena
Arena Maths	1504#12	–	–	1491#27	Arena (LMArena)

Knowledge

Benchmark	no reasoning	xhigh	max	not stated	Source
SimpleQA Verified	–	51.7%#25	–	–	Epoch AI Benchmarking Hub
LiveBench 1.0 Data Analysisnot in index	–	78.3%#25	–	–	LiveBench
AA-Omniscience	14.8#82	–	27.3#41	–	Artificial Analysis
MMLU-Proeffort not stated	–	–	–	89.9%#8	Vals AI
LegalBencheffort not stated	–	–	–	85.3%#24	Vals AI
CorpFineffort not stated	–	–	–	66.1%#22	Vals AI
TaxEvaleffort not stated	–	–	–	75.3%#20	Vals AI
CaseLaw v2not in indexeffort not stated	–	–	–	68.4%#5	Vals AI
Excel Modeling Benchmarknot in indexeffort not stated	–	–	–	63.7%#26	Vals AI
MedCodenot in indexeffort not stated	–	–	–	54.9%#7	Vals AI
MedScribenot in indexeffort not stated	–	–	–	83.0%#47	Vals AI
EBR-benchnot in index	–	–	19.1%#17	–	Epoch AI Benchmarking Hub
GDP.pdfnot in index	–	–	21.0%#30	–	Epoch AI Benchmarking Hub
AA-Omniscience: accuracynot in index	44.7%#91	–	48.9%#67	–	Artificial Analysis
AA-Omniscience: non-hallucinationnot in index	45.9%#139	–	57.7%#85	–	Artificial Analysis

Instruction following

Benchmark	no reasoning	high	xhigh	max	not stated	Source
LiveBench Languagenot in index	–	–	77.9%#40	–	–	LiveBench
LiveBench Instruction Followingnot in index	–	–	66.7%#39	–	–	LiveBench
IFBench	43.6%#219	–	–	58.6%#129	–	Artificial Analysis
Arena Instruction Followingnot in index	–	1502#5	–	–	1493#10	Arena (LMArena)

Human preference

Benchmark	high	not stated	Source
Arena Text	1501#7	1494#11	Arena (LMArena)
EQ-Bench 4effort not stated	–	1311#5	EQ-Bench
Arena Creative Writingnot in index	1489#6	1482#11	Arena (LMArena)
Arena Multi-turnnot in index	1516#5	1514#7	Arena (LMArena)
Arena Longer Queriesnot in index	1515#6	1508#9	Arena (LMArena)

Multimodal

Benchmark	no reasoning	high	max	not stated	Source
Arena Vision	–	1313#9	–	1317#5	Arena (LMArena)
MMMU-Pro	76.4%#88	–	78.8%#67	–	Artificial Analysis
MMMU-Pro (Vals)not in indexeffort not stated	–	–	–	85.5%#24	Vals AI
MortgageTaxnot in indexeffort not stated	–	–	–	70.3%#3	Vals AI
SAGEnot in indexeffort not stated	–	–	–	56.1%#1	Vals AI
Blueprint-Bench 2not in indexeffort not stated	–	–	–	24.5%#24	Epoch AI Benchmarking Hub

Long context

Benchmark	no reasoning	high	max	not stated	Source
AA-LCR	75.7%#153	–	78.7%#109	–	Artificial Analysis
Arena Documentnot in index	–	1493#5	–	1494#3	Arena (LMArena)
MLCRnot in index	28.9%#21	–	–	–	Artificial Analysis

Composite

Benchmark	no reasoning	xhigh	max	not stated	Source
Epoch Capabilities Indexnot in indexeffort not stated	–	–	–	156.8#26	Epoch AI Benchmarking Hub
LiveBench 1.0not in index	–	76.5%#29	–	–	LiveBench
AA Intelligence Index v4.3.2	30.9#115	–	40.7#61	–	Artificial Analysis
Vals Multimodal Indexnot in indexeffort not stated	–	–	–	67.4%#10	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 Vertex	70 tok/s	1.10 s	$5.00	$25.00	1M	–
Amazon Bedrock	61 tok/s	1.25 s	$5.00	$25.00	1M	–
Claude Platform on AWS	51 tok/s	0.92 s	$5.00	$25.00	1M	–
Anthropic	40 tok/s	0.75 s	$5.00	$25.00	1M	–

Price history

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

input outputnow $5.50 in · $27.50 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 $0.500 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0095	$0.0081	6.2 s
Summarize a 30-page report	12,000 / 600	$0.075	$0.035	11.2 s
Code edit	6,000 / 1,500	$0.068	$0.047	26.2 s
Agentic coding session	60,000 / 4,000	$0.400	$0.198	1.1 min
Structured extraction	2,000 / 200	$0.015	$0.0083	4.6 s