LLMcompare

Search

Search for a command to run...

A · closed · Claude Opus

Claude Opus 4.8

Anthropic

B · closed · Claude Opus

Claude Opus 5

Anthropic

0:17

overall capability wins (2 ties across 19 scored). Claude Opus 5 leads overall.

  • Reasoning0:5
  • Coding0:8
  • Arena0:1
  • Agents0:2
  • Tool use0:1
  • Composite indices0:2

Verdict

Claude Opus 5 leads coding (widest gap: +148.2 on WebDev Elo); Claude Opus 5 ranks higher on LMArena (+20 Elo).

  • Claude Opus 5 leads coding (widest gap: +148.2 on WebDev Elo)
  • Claude Opus 5 ranks higher on LMArena (+20 Elo)

Cost for 1M in + 250K out

Illustrative chat workload at primary-provider list prices.

Claude Opus 4.8
$11.25
Claude Opus 5
$11.25

Price per 1M tokens

Full comparison

Organization

Anthropic

Anthropic

Family

Claude Opus

Claude Opus

License

Proprietary

Proprietary

Open weights

No

No

Release

May 28, 2026

Jul 24, 2026

Knowledge cutoff

-

-

API / provider

Anthropic

Anthropic

Modalities

text, image → text

text, image → text

Specs

Context window

1M

1M

tie

Max output

128K

128K

tie

Parameters

—

—

—

Pricing

Input $/1M

$5

$5

tie

Output $/1M

$25

$25

tie

Blended $/1M (3∶1)

$10

$10

tie

Speed

tok/s

72

72

tie

TTFT (s)

0.55

0.55

tie

Reasoning

MMLU-Pro

—

—

—

GPQA Diamond

92.0

93.2

B +1.2 pts

Humanity's Last Exam

48.7

54.9

B +6.2 pts

AIME 2025

—

—

—

MATH-500

—

—

—

Humanity's Last Exam (with tools)

—

—

—

AA-Omniscience Accuracy

48.8

60.9

B +12.1 pts

AA-LCR v1.1

77.7

79.3

B +1.6 pts

CritPt

20.9

29.1

B +8.2 pts

MMMU-Pro

—

84.7

—

IFBench

62.2

—

—

Chartography

—

—

—

Chartography (With Tools)

—

—

—

Coding

SWE-bench Verified

—

—

—

SWE-bench Pro

69.2

—

—

SWE-bench Multilingual

—

—

—

LiveCodeBench

—

—

—

Terminal-Bench 2.1

84.6

89.1

B +4.5 pts

Aider Polyglot

—

—

—

Terminal-Bench 3

21.1

42.7

B +21.6 pts

BigCodeBench

—

—

—

SciCode

54.4

56.4

B +2.0 pts

CursorBench

62.3

70.0

B +7.7 pts

SWE-Rebench

56.5

63.4

B +6.9 pts

NL2Repo-Bench

—

—

—

DeepSWE

59.0

73.7

B +14.7 pts

WebDev Arena

1,539

1,687

B +148 Elo

Terminal-Bench 4.0

23.6

51.8

B +28.2 pts

CursorBench 4.0

—

—

—

FrontierCode v1.1 (Main)

—

—

—

Arena

LMArena Elo

1,473

1,493

B +20 Elo

Claude Opus 4.8

Context
1M / 128K out
Parameters
—
Price
$5 / $25
Speed
72 tok/s · 0.55s TTFT
Modalities
text, image → text
License
Proprietary

Previous Opus flagship with 1M context; still competitive for coding agents and computer use.

Claude Opus 5

Context
1M / 128K out
Parameters
—
Price
$5 / $25
Speed
72 tok/s · 0.55s TTFT
Modalities
text, image → text
License
Proprietary

Anthropic's July 2026 flagship - top Artificial Analysis Intelligence Index and strong agentic coding at Opus 4.8 pricing.

Benchmark charts

Winner bars are emphasized. Per-benchmark deltas sit above each chart.

Reasoning & knowledge

GPQA: B +1.2HLE: B +6.2CritPt: B +8.2AA-LCR: B +1.6Omniscience Acc.: B +12.1

Coding

TermBench: B +4.5TermBench 3: B +21.6TermBench 4: B +28.2SciCode: B +2.0CursorBench: B +7.7SWE-Rebench: B +6.9DeepSWE: B +14.7WebDev Elo: B +148

Tool use & function calling

τ³-Banking: B +7.9

Agents & computer use

TB-Science: B +19.5GDPval-AA: B +246

Arena

Arena Elo: B +20

Composite indices

AA Index: B +12.1AA Index v4.3: B +8.7

Pick a different pair · Back to catalog