VerdictPal
Join

Models / Anthropic

Claude Opus 5 (Adaptive Reasoning, Max Effort)

Claude Opus · Released 2026-07-24

Anthropic's July 2026 Opus line. AA Intelligence Index v4.3 is 50.7 at max effort — behind Fable 5.1 (53.4) and GPT-6 Astra (52.8). API list $5 / $25 per 1M tokens. Vals Index still 67.21%.

Frontier model tracked on Artificial Analysis. Use the benchmark ledger below for sourced scores; we do not infer numbers beyond published rows.

#3 of 4 on Vals Index · current snapshot
67 Vals Index · 16 benchmark rows · 5 sourcesPosition is benchmark-specific — not a cross-family or cross-source ranking.
Composite62
Context1M
Input / 1M$5.0
Output / 1M$25
Knowledge cutoffopen
Output speed55 t/s
TTFT (default API)33.87s
Statussolid

Family profile

Best published score in each covered benchmark family.

80/100 avg
Knowledge
61
Reasoning
93
Coding
74
Agentic
89
Multimodal
85
Long context
79

6 tested benchmark families

Benchmark placements

Where Claude Opus 5 (Adaptive Reasoning, Max Effort) places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.

BenchmarkFamilyRankScoreSourceDate
Terminal-BenchAgentic#2/ 18989Artificial Analysis2026-09-10
Artificial Analysis Intelligence IndexReasoning#3/ 18851Artificial Analysis2026-09-10
Humanity's Last ExamReasoning#3/ 18555Artificial Analysis2026-09-10
AA time to first tokenPerformance#10/ 190Artificial Analysis2026-09-10
GPQA DiamondReasoning#14/ 19093Artificial Analysis2026-09-10
SciCodeCoding#14/ 9456Artificial Analysis2026-09-10
τ³-Bench BankingAgentic#15/ 11142Artificial Analysis2026-09-10
AA-LCRLong context#44/ 18279Artificial Analysis2026-09-10
AA output speedPerformance#69/ 190Artificial Analysis2026-09-10
Family context

The three highest-scoring models with pages in each capability family. Where Claude Opus 5 (Adaptive Reasoning, Max Effort) shows up, it's highlighted.

KnowledgeClaude Opus 5 (Adaptive Reasoning, Max Effort) · 61
  1. 1
    GPT 5.5
    91
  2. 2
    Claude Opus 4.8
    91
  3. 3
    Claude Sonnet 4.6
    91
ReasoningClaude Opus 5 (Adaptive Reasoning, Max Effort) · 93
  1. 1
    GPT-6 Astra
    96
  2. 2
    Claude Mythos Preview
    95
  3. 3
    Gemini 3.7 Flash
    95
CodingClaude Opus 5 (Adaptive Reasoning, Max Effort) · 74
  1. 1
    GPT 5.5
    96
  2. 2
    Claude Opus 4.8
    96
  3. 3
    Claude Opus 4.6
    96
AgenticClaude Opus 5 (Adaptive Reasoning, Max Effort) · 89
  1. 1
    GPT 5.5
    100
  2. 2
    Claude Opus 4.8
    100
  3. 3
    Claude Fable 5
    100
MultimodalClaude Opus 5 (Adaptive Reasoning, Max Effort) · 85
  1. 1
    Gemini 3.7 Flash
    86
  2. 2
    Claude Opus 5 (Adaptive Reasoning, Max Effort)
    85
  3. 3
    Gemini 3.5 Flash
    84
Long contextClaude Opus 5 (Adaptive Reasoning, Max Effort) · 79
  1. 1
    Kimi K3
    89
  2. 2
    Claude Fable 5.1
    85
  3. 3
    GPT-5.5 (high)
    84
Newest receipts

Benchmark rows added to the public ledger for Claude Opus 5 (Adaptive Reasoning, Max Effort) in the last 120 days. Older rows live in the full table below.

16 recent
BenchmarkFamilyScoreSourceDays ago
AA-LCRLong context79.33%Artificial Analysis4
AA output speedPerformance55 t/sArtificial Analysis4
AA time to first tokenPerformance33.87sArtificial Analysis4
Artificial Analysis Intelligence IndexReasoning50.7%Artificial Analysis4
GPQA DiamondReasoning93.2%Artificial Analysis4
Humanity's Last ExamReasoning54.9%Artificial Analysis4
SciCodeCoding56.4%Artificial Analysis4
τ³-Bench BankingAgentic42.06%Artificial Analysis4
Terminal-BenchAgentic89.14%Artificial Analysis4
Vals IndexAgentic67.21%Vals AI — Vals Index13
CritPtReasoning29.1%Artificial Analysis — CritPt13
AA-OmniscienceKnowledge60.9%Artificial Analysis — AA-Omniscience13
MMMU-ProMultimodal84.7%Artificial Analysis — MMMU-Pro13
Terminal-Bench-ScienceAgentic30%Terminal-Bench-Science 0.1 announcement18
DeepSWECoding74%DeepSWE leaderboard41
Vending-Bench 2Agentic11181.87Andon Labs Vending-Bench 241
Family coverage

A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.

Knowledge61
Reasoning93
Coding74
Agentic89
Multimodal85
Long context79
Price vs. performance
0255075100$0.10$0.30$1$3$10$30input USD / 1M tokens · log scale →← intelligence

Hover any dot for name, score, and input price. Keyboard: tab through the top twelve, or use the ranking below.

AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMetaxAIMiniMaxZhipuXiaomiOtherMeituanCohereNVIDIA

Full benchmark ledger

Every catalog benchmark for Claude Opus 5 (Adaptive Reasoning, Max Effort). Scores link to the original source; gaps mean no public row exists yet.

16 sourced rows

16 of 37 catalog benchmarks have a sourced row for Claude Opus 5 (Adaptive Reasoning, Max Effort).

Knowledge

Reasoning

Coding

Agentic

Multimodal

Long context

Performance

Sources