BenchLeader

Claude Opus 4

Reasoning effort
Quality#103 of 333
55.2Index
Above the median
+5.2 on the median
Price#242 of 249
$30.00per 1M tokens
Dearest quarter
35× the median

90 cheaper ranked models score higher: Claude Opus 5.5 (69.4, $8.00), Gemini 4 Carbon (69.3, $8.00), GPT-6 Astra (68.3, $20.00) and 87 more.

As of 8 Nov 2026, Claude Opus 4 by Anthropic scores 55.2 on the BenchLeader Index (593 ranked configurations), with its best results in maths (65) and weakest in reasoning (48). API pricing: $15.00 per 1M input tokens and $75.00 per 1M output tokens ($30.00 blended at 3:1), list price on 8 Nov 2026. Released 22 May 2025.

Key facts (as of 8 Nov 2026)

  • Claude Opus 4 API pricing: $15.00 per 1M input tokens and $75.00 per 1M output tokens ($30.00 blended at 3:1) (list price, 8 Nov 2026).
  • Claude Opus 4 was released on 22 May 2025 (Anthropic).

In detail. Claude Opus 4 is an Anthropic proprietary reasoning model, released 22 May 2025. Its best configuration (thinking reasoning effort) ranks #206 of 593 on the BenchLeader Index at 55.2 ±4.1, in the upper half. The ± is the point: 269 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 maths (65) and lowest in reasoning (48). At $30.00 per million tokens blended it is among the most expensive ranked models. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 6 Nov 2026.

Blended price
$30.00/M
$15.00 in · $75.00 out
via Artificial Analysis
Output speed
–
First answer
–
Context
200k
via models.dev
Full answer
–
Cost per task
–
Tokens per task
–
Released
22 May 2025
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index55
  2. Reasoning48
  3. Coding54
  4. Agents & tools61
  5. Maths65
  6. Knowledge49
  7. Instruction following54
  8. Human preference61
  9. Multimodal55
  10. Long context59

Get an email when Anthropic ships a new model

And once when it gets ranked, plus the weekly review on Mondays.

Choose providers and emails

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.

ModelReleasedIndexRank
Claude Opus 5.5max22 Sep 202669.4#1
Claude Opus 5max24 Jul 202665.4#15
Claude Opus 4.8high28 May 202660.6#74
Claude Opus 4.716 Apr 202663.2#40
Claude Opus 4.6high5 Feb 202662.7#44
Claude Opus 4.5thinking24 Nov 202559.7#93
Claude Opus 4.1thinking5 Aug 202557.9#140
Claude Opus 4thinkingthis page22 May 202555.2#206
Claude 3 Opus4 Mar 202439.0#537

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.

EffortIndexRankSpeedFirst answerChat reply costCategories
no reasoning––––$0.029Instruction following 46
thinkingbest55.2#206––$0.029Agents & tools 61 · Coding 54 · Human preference 61 · Instruction following 54 · Knowledge 49 · Long context 59 · Maths 65 · Multimodal 55 · Reasoning 48
not stated52.6#266––$0.029Agents & tools 56 · Coding 53 · Human preference 59 · Instruction following 58 · Knowledge 55 · Long context 37 · Maths 57 · Multimodal 53 · Reasoning 47

How its index has moved

8 Sep 2026 to 29 Oct 2026
475461
  • thinking
  • not stated

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

Benchmarkno reasoningthinkingnot statedSourceTrend
GPQA Diamondeffort not stated––76.3%#152Epoch AI Benchmarking Hub
Humanity's Last Exameffort not stated––10.7%#30Scale AI / CAIS
SimpleBencheffort not stated––58.8%#31SimpleBench
Arena Hard Prompts–1457#1061436#133Arena (LMArena)
GPQA Diamond (AA)not in index70.1%#30179.6%#200–Artificial Analysis
Humanity's Last Exam (AA)not in index6.2%#41912.3%#290–Artificial Analysis
GPQA Diamond (Vals)not in indexeffort not stated––71.7%#89Vals AI
Kagi LLM Benchmark–74.3%#1559.6%#52Kagi LLM Benchmark
GPQA Diamond (HELM)not in index–70.9%#966.6%#15HELM Capabilities
ARC-AGI-1–35.7%#20122.5%#219ARC Prize
ARC-AGI-2–8.6%#1661.3%#221ARC Prize
LMCAnot in indexeffort not stated––37.4%#119Epoch AI Benchmarking Hub
DTBenchnot in indexeffort not stated––81.6%#126Epoch AI Benchmarking Hub

Coding

Benchmarkno reasoningthinkingnot statedSourceTrend
SWE-bench Verified (Epoch)effort not stated––70.7%#25Epoch AI Benchmarking Hub
WeirdMLeffort not stated––43.7%#86WeirdML
GSO-Bencheffort not stated––6.9%#23GSO-Bench
Arena Coding–1498#831464#126Arena (LMArena)
LiveCodeBench–70.2%#9062.6%#102Vals AI
Aider Polygloteffort not stated––72.0%#8Aider polyglot leaderboard
SWE-bench Verified (bash only)effort not stated––67.6%#16SWE-bench
SWE-bench Verified (any scaffold)not in indexeffort not stated––73.2%#9SWE-bench
AlgoTunenot in indexeffort not stated––1.3#17AlgoTune
Design Arena Websitenot in indexeffort not stated––1162#114Design Arena
LiveCodeBench (AA)not in index54.2%#12863.6%#99–Artificial Analysis

Agents & tools

Benchmarkthinkingnot statedSourceTrend
Cybencheffort not stated–38.0%#7Cybench
Terminal-Bench Hard31.1%#109–Artificial Analysis
τ²-Bench Telecom (AA)not in index73.4%#147–Artificial Analysis
Arena Searchnot in indexeffort not stated–1152#35Arena (LMArena)
DeepResearch Benchnot in indexeffort not stated–46.8%#26DeepResearch Bench
METR Time Horizonsnot in indexeffort not stated–63.9%#16METR
BrowseComp-Plusnot in indexeffort not stated–15.5%#12BrowseComp-Plus
BrowseCompnot in indexeffort not stated–1.0%#18OpenAI

Maths

Benchmarkthinkingnot statedSourceTrend
OTIS Mock AIMEeffort not stated–64.4%#165Epoch AI Benchmarking Hub
MATH Level 5effort not stated–85.0%#27Epoch AI Benchmarking Hub
AIME (Vals)effort not stated–41.3%#68Vals AI
MGSMeffort not stated–93.8%#8Vals AI
Omni-MATH (HELM)61.6%#851.1%#18HELM Capabilities
MATH 500not in indexeffort not stated–90.4%#24Vals AI
Arena Maths1422#1241403#149Arena (LMArena)

Knowledge

Benchmarkthinkingnot statedSourceTrend
MMLU-Proeffort not stated–86.2%#47Vals AI
LegalBencheffort not stated–83.1%#60Vals AI
TaxEvaleffort not stated–71.9%#69Vals AI
MedQAeffort not stated–92.9%#26Vals AI
MultiNRC33.9%#2229.0%#24Scale AI SEAL
MMLU-Pro (HELM)not in index87.5%#285.9%#6HELM Capabilities

Instruction following

Benchmarkno reasoningthinkingnot statedSourceTrend
IFBench43.3%#22853.7%#156–Artificial Analysis
MultiChallenge–58.6%#19–Scale AI SEAL
TutorBench–49.7%#1945.5%#25Scale AI SEAL
IFEval (HELM)–84.9%#2191.8%#7HELM Capabilities
WildBench (HELM)–85.2%#1283.3%#18HELM Capabilities
Arena Instruction Followingnot in index–1444#741415#120Arena (LMArena)

Human preference

Benchmarkthinkingnot statedSourceTrend
Arena Text1426#1181414#139Arena (LMArena)
Arena Creative Writingnot in index1433#671417#87Arena (LMArena)
Arena Multi-turnnot in index1439#1061431#116Arena (LMArena)
Arena Longer Queriesnot in index1466#661448#93Arena (LMArena)
Lech Mazur Creative Writingnot in indexeffort not stated–83.6%#7Lech Mazur

Multimodal

Benchmarkthinkingnot statedSourceTrend
Arena Vision1192#971175#105Arena (LMArena)
VISTA47.0%#1843.5%#32Scale AI SEAL
MMMU (validation)effort not stated–76.5%#13MMMU
MMMU-Pro (Vals)not in indexeffort not stated–73.3%#60Vals AI
MortgageTaxnot in indexeffort not stated–58.6%#68Vals AI
GeoBenchnot in indexeffort not stated–49.0%#23GeoBench
VPCTnot in indexeffort not stated–38.0%#21Epoch AI Benchmarking Hub

Long context

Benchmarkthinkingnot statedSourceTrend
Fiction.LiveBench 120keffort not stated–37.5%#27Fiction.live
AA-LCR v1.169.3%#235–Artificial Analysis

Composite

Benchmarkno reasoningthinkingnot statedSourceTrend
Epoch Capabilities Indexnot in indexeffort not stated––142.8#99Epoch AI Benchmarking Hub
AA Intelligence Index v4.3.2not in index16.6#28320.6#235–Artificial Analysis
AA Progress Index v1not in index200.6#412233.9#306–Artificial Analysis
HELM Capabilities meannot in index–78.0%#875.7%#13HELM Capabilities

Safety & honesty

Benchmarkthinkingnot statedSourceTrend
FORTRESSnot in index24.8%#3827.6%#30Scale AI SEAL
MASKnot in indexpossible contamination87.9%#1080.3%#24Scale AI SEAL

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 $1.50 per 1M. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.029$0.025–
Summarize a 30-page report12,000 / 600$0.225$0.103–
Code edit6,000 / 1,500$0.203$0.142–
Agentic coding session60,000 / 4,000$1.20$0.593–
Structured extraction2,000 / 200$0.045$0.025–

More from Anthropic

New this month

See also

Embed or cite

Show where Claude Opus 4 ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page. To see ranks like this on OpenRouter and Hugging Face model pages, get the Chrome extension.

[![Claude Opus 4 on BenchLeader: #102 overall](https://www.benchleader.com/badge/claude-opus-4.svg)](https://www.benchleader.com/models/claude-opus-4)

Cite this

The ranking changes daily, so a reference carries the data date. APA style:

BenchLeader. (2026, November 8). Claude Opus 4 on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 8, 2026, from https://www.benchleader.com/models/claude-opus-4

BibTeX:

@misc{benchleader_claude_opus_4_20261108,
  author       = {{BenchLeader}},
  title        = {{Claude Opus 4 on the BenchLeader Index (method v1.4)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/claude-opus-4}},
  note         = {Data as of 2026-11-08}
}

Every badge style, the embeddable top-10 widget and how to cite the Index

Questions people ask

What are Claude Opus 4's benchmark scores?
As of 8 Nov 2026, Claude Opus 4 (thinking effort) scores 55.2 ±4.1 on the BenchLeader Index, strongest in maths (65) and weakest in reasoning (48). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does Claude Opus 4 rank?
Claude Opus 4 (thinking effort) is #206 of 593 ranked configurations on 8 Nov 2026. 269 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 69.4.
How much does Claude Opus 4 cost per 1M tokens?
As of 8 Nov 2026, Claude Opus 4 costs $15.00 per 1M input tokens and $75.00 per 1M output tokens, $30.00 blended (three input tokens per output token).
When was Claude Opus 4 released?
Anthropic released Claude Opus 4 on 22 May 2025. BenchLeader has results for it at 3 reasoning-effort settings.

Data as of 8 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “Claude Opus 4: benchmarks, pricing, speed and rank”, data as of 8 Nov 2026, https://www.benchleader.com/models/claude-opus-4.