BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases12
AnthropicCompare with another model

Claude 3 Opus

Claude 3 Opus is an Anthropic proprietary model, released 4 Mar 2024. Its best configuration ranks #691 of 785 on the BenchLeader Index at 39.2 ±4.2, in the lower half. The ± is the point: 239 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 (48) and lowest in knowledge (26). At $30.00 per million tokens blended it is among the most expensive ranked models. Last measured 22 Oct 2026.

Blended price
$30.00/M
$15.00 in · $75.00 out
Output speed
–
First answer
–
Context
200k
Full answer
–
Cost per run
–
Released
4 Mar 2024
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index39
  2. Reasoning36
  3. Coding38
  4. Agents & tools38
  5. Maths41
  6. Knowledge26
  7. Human preference48
  8. Multimodal31
  9. Composite39

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 Sept 202672.6#1
Claude Opus 5max24 Jul 202668.8#15
Claude Opus 4.8max28 May 202663.3#67
Claude Opus 4.716 Apr 202663.3#66
Claude Opus 4.6high5 Feb 202662.7#78
Claude Opus 4.5thinking24 Nov 202559.9#119
Claude Opus 4.1thinking5 Aug 202556.4#204
Claude Opus 4thinking22 May 202555.0#236
Claude 3 Opusthis page4 Mar 202439.2#691

How its index has moved

8 Sept 2026 to 19 Oct 2026
353944

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

Benchmarknot statedSourceTrend
GPQA Diamond47.2%#230Epoch AI Benchmarking Hub
SimpleBench23.5%#75SimpleBench
Arena Hard Prompts1330#232Arena (LMArena)
GPQA Diamond (AA)not in index48.9%#435Artificial Analysis
Humanity's Last Exam (AA)not in index2.8%#610Artificial Analysis
Chess Puzzlesnot in index3.0%#135Epoch AI Benchmarking Hub
LMCAnot in index17.0%#167Epoch AI Benchmarking Hub
DTBenchnot in index61.6%#178Epoch AI Benchmarking Hub
ForecastBenchnot in index62.4%#56ForecastBench
WinoGrandenot in index88.5%#2Epoch AI Benchmarking Hub

Coding

Benchmarknot statedSourceTrend
WeirdML19.2%#142WeirdML
Arena Coding1355#238Arena (LMArena)
SWE-bench Verified (any scaffold)not in index15.8%#59SWE-bench

Agents & tools

Benchmarknot statedSourceTrend
Cybench10.0%#15Cybench
METR Time Horizonsnot in index29.5%#36METR

Maths

Benchmarknot statedSourceTrend
OTIS Mock AIME4.7%#246Epoch AI Benchmarking Hub
MATH Level 537.5%#66Epoch AI Benchmarking Hub
Arena Maths1312#217Arena (LMArena)

Knowledge

Benchmarknot statedSourceTrend
SimpleQA Verified12.6%#82Epoch AI Benchmarking Hub
MMLUnot in index84.6%#10Epoch AI Benchmarking Hub

Instruction following

Benchmarknot statedSourceTrend
Arena Instruction Followingnot in index1314#224Arena (LMArena)

Human preference

Benchmarknot statedSourceTrend
Arena Text1322#232Arena (LMArena)
Arena Creative Writingnot in index1288#229Arena (LMArena)
Arena Multi-turnnot in index1323#223Arena (LMArena)
Arena Longer Queriesnot in index1330#227Arena (LMArena)

Multimodal

Benchmarknot statedSourceTrend
Arena Vision1023#137Arena (LMArena)
VISTA27.8%#50Scale AI SEAL
MMMU (validation)59.4%#34MMMU

Composite

Benchmarknot statedSourceTrend
Epoch Capabilities Indexnot in index126.9#158Epoch AI Benchmarking Hub
AA Intelligence Index v4.3.28.7#430Artificial Analysis

Safety & honesty

Benchmarknot statedSourceTrend
MASKnot in index79.0%#26Scale 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. Reasoning tokens are not modelled.

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

More from Anthropic

New this month

See also

Best for codingAll Anthropic models

Questions people ask

What are Claude 3 Opus's benchmark scores?
As of 24 Oct 2026, Claude 3 Opus scores 39.2 ±4.2 on the BenchLeader Index, strongest in human preference (48) and weakest in knowledge (26). That rests on 15 independent benchmark results, and every score on this page links to its publisher.
How does Claude 3 Opus rank?
Claude 3 Opus is #691 of 785 ranked configurations on 24 Oct 2026. 239 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 72.6.
How much does Claude 3 Opus cost per 1M tokens?
As of 24 Oct 2026, Claude 3 Opus 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 3 Opus released?
Anthropic released Claude 3 Opus on 4 Mar 2024.

Data as of 24 Oct 2026. Compare with another model.

Cite as: BenchLeader, “Claude 3 Opus: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/claude-3-opus.