The Model Index
Models/Anthropic

Claude Opus 5.5

AnthropicClosed weightsFlagship

Claude Opus 5.5 is a closed AI model from Anthropic with an index score of 195 (GPT-4 = 100), rank 5 of 217.

195178–211Index score#5 of 217 models
$4.0/ $20Price per million tokensinput / output · via OpenRouter
1MContext windowtokens it reads at once
22 Sep 2026Released2 months ago

Where it sitson the frontier

801001201401601802002023202420252026GPT-4 at launch = 100Claude Opus 5.5

Among the first models to reach this score.

Test resultsby area · tick = best by any model

Reasoning & knowledgeevidence
GPQA Diamond (Epoch AI)
94%
Humanity's Last Exam
no result
Mathsevidence
FrontierMath (tiers 1-3)
91%
Codingevidence
SWE-bench Verified
no result
Terminal-Bench 4.0
65%
Agentsevidence
APEX-Agents
74%
Professional workevidence
Finance Agent (v2)
59%
Novel problemsevidence
ARC-AGI-2
93%
Long tasksno evidence yet
METR time horizon
no result

Third-party results only, at the model’s best standard setting. Hover a bar for who ran it.

What the lab says2 of 9 also run by third parties

TestAnthropic saysThird parties
Terminal-Bench 4.0Claude Code harness, 5 tries per task. Anthropic gives a margin of ±2.6 points.66.4%65.1%Vals AI
Terminal-Bench Science 0.1Claude Code harness, 3 tries per task. Anthropic gives a margin of ±3.5–5 points.58.7%47.1%Vals AI
FrontierCode v1.1 (Main)54.4%not run yet
CursorBench 4.057.8%not run yet
GDPval-AA v2.11846 Elonot run yet
AutomationBench40.0%not run yet
Humanity's Last Examwith tools67.7%not run yet
OSWorld 2.1partial credit81.8%not run yet
Chartographywith tools89.0%not run yet

From Anthropic’s launch post, Introducing Claude Opus 5.5, copied from the table in the post. Shown, never scored; settings can differ from the third-party run.

What others sayrank on 12 leaderboards

Best attop three among current models

Where to use itprices checked 18 Nov 2026

Host prices from OpenRouter; going direct can differ.

Facts and sources

Released22 Sep 2026
Announcementanthropic.com ↗
Release entryClaude Opus 5.5
Scored asClaude Opus 5.5
Output speed–
Hugging Faceclosed weights

Anthropic releasesaround this one

This page as a chartDownload PNG