Orivel Orivel
Open menu

Claude Opus 5

Explore benchmark scores, genre strengths, weaknesses, and recent examples for Claude Opus 5 on Orivel.

This page shows results from the earlier method (one-on-one matches between two models). Current results are in the snapshot rankings.

See current results →

Model Overview

Provider: Anthropic · claude-opus-5 Retired

Released

2026-07-24

Context

1M tokens

Input

$5.00 / 1M

Output

$25.00 / 1M

Max output

128k tokens

Retired from new Orivel comparisons on September 24, 2026, when Claude Opus 5.5 took the Anthropic balanced slot at a lower price ($4 / $20 per 1M tokens against $5 / $25). Past results stay published.

Claude Opus 5, released July 24, 2026, is Anthropic's model for complex agentic coding and enterprise work. On Orivel it takes the Anthropic balanced slot from Opus 4.8 at the identical price, sitting below Claude Fable 5, which remains the frontier tier. Anthropic positions it as near-Fable 5 intelligence at Opus speed and cost.

Benchmarks

Measure Opus 5 Note
SWE-bench Verified 96.0% averaged over five trials
SWE-bench Multilingual 89.5%
SWE-bench Pro 79.2%
Frontier-Bench v0.1 — more than double Opus 4.8
ARC-AGI 3 — about 3x the next-best competitor
OSWorld 2.0 — beats Fable 5's best at one third the cost

Specs and pricing

Item Opus 5
Context window 1M tokens
Max output 128k tokens
Knowledge cutoff May 2026 — the most recent of any current Claude model
Input / 1M $5 (unchanged from Opus 4.8)
Output / 1M $25 (unchanged from Opus 4.8)
Adaptive thinking On by default — a change from Opus 4.8
Sampling parameters temperature and similar are no longer accepted

Why this is the default we'd pick

This is the model we would default to for quality work. It arrived at exactly Opus 4.8's price while clearing a large gap in capability, which is the rarest kind of release — an upgrade that costs nothing to take. On our own head-to-head sessions it wins the large majority of its matchups, and it takes first place more often than any other model in its tier.

The case for spending five times more on Fable 5.1 instead is narrower than the marketing suggests. Unless the task is genuinely at the edge of what these models can do, the extra money buys less than the price difference implies.

Retirement notes

  • Released July 24, 2026 for complex agentic coding and enterprise work
  • SWE-bench Verified 96.0% (5-trial average); SWE-bench Pro 79.2%; SWE-bench Multilingual 89.5%
  • Frontier-Bench v0.1: more than double Opus 4.8's score; ARC-AGI 3: ~3x the next-best competitor
  • Beats Fable 5's best OSWorld 2.0 result at one third of the cost
  • Gains on scientific research: +10.2 points on organic chemistry, +7.7 points on protein prediction vs Opus 4.8
  • 1M-token context window; up to 128k output tokens
  • Adaptive thinking ON by default (Opus 4.8 required opting in); temperature / top_p / top_k no longer accepted
  • Full effort range including `max`; fast mode available at 2.5x speed
  • Pricing unchanged from Opus 4.8: $5 input / $25 output per 1M tokens
  • Knowledge cutoff: May 2026
Official announcement

Specs across the current lineup

Published specs and pricing, comparable before any benchmark data exists.

Model Context Max output Input / 1M Output / 1M
openai GPT-6 Astra 1M 128k $10.00 $50.00
google Gemini 3.8 Flash 1M 66k $0.75 $3.75
anthropic Claude Fable 5.1 1M 128k $10.00 $50.00
anthropic Claude Opus 5.5 1M 128k $4.00 $20.00
google Gemini 3.5 Flash-Lite 1M 66k $0.30 $2.50
openai GPT-6.1 Sol 1M 128k $2.00 $10.00
openai GPT-6 Luna 1M 128k $0.10 $0.50
anthropic Claude Sonnet 5.5 1M 128k $2.00 $10.00

Overall Performance

Overall Rank

-

Overall win rate

97%

Average Score

86

Wins

31

Sample Count

32

Win Rate by Model

Compare by Genre

Strength by Evaluation Criteria

Average score by criterion (out of 10)

Completeness 9.33 6 samples
Trade-off Reasoning 9.27 3 samples
Scalability & Reliability 9.23 3 samples
Helpfulness 9.20 3 samples
Specificity 9.17 3 samples
Instruction Following 9.16 9 samples
Reasoning Quality 9.10 3 samples
Architecture Quality 9.10 3 samples
Depth 9.07 3 samples
Empathy 8.90 3 samples
Style Quality 8.70 3 samples
Usefulness 8.70 3 samples

Latest Tasks

Explanation

Anthropic Claude Opus 5 VS OpenAI GPT-6 Astra

Explaining Eventual Consistency to a Relational Database Developer

Explain the concept of 'eventual consistency' in distributed systems to a junior software developer who is only familiar with traditional relational databases a...

288
Sep 9, 2026 01:12

Analysis

OpenAI GPT-5.5 VS Anthropic Claude Opus 5

Analyze Marketing Strategies for an Eco-Friendly Coffee Shop

Analyze the three marketing strategies described in the context and recommend the best one for 'The Daily Grind' coffee shop to launch their new line of reusabl...

349
Aug 21, 2026 09:39

Roleplay

Google Gemini 2.5 Flash-Lite VS Anthropic Claude Opus 5

A Restaurant Manager Responds to a Severe Allergy Request

Respond in character as Elena, the restaurant manager, to the customer's message below. Write one natural reply of 130–190 words. Be warm, calm, and direct rath...

391
Aug 19, 2026 09:38

Idea Generation

OpenAI GPT-5.5 VS Anthropic Claude Opus 5

Reimagining Urban Community Spaces

Generate a list of 5 distinct, innovative, and practical ideas for a new type of community space. The space is a vacant, 200-square-meter ground-floor retail un...

512
Aug 1, 2026 09:38

Summarization

Google Gemini 2.5 Flash VS Anthropic Claude Opus 5

Summarize the Rellan Rainwater Reuse Pilot

Write a standalone summary of the fictional source passage below in 220 to 280 words. Use continuous prose rather than a list. Your summary must explain why the...

501
Jul 31, 2026 09:41

Creative Writing

Google Gemini 2.5 Pro VS Anthropic Claude Opus 5

The Last Item in Lost Property

Write a complete short story of 700–1,000 words for a general adult audience. The story takes place during the final hour before a city train station’s lost-pro...

531
Jul 29, 2026 09:40

Humor

Google Gemini 2.5 Flash-Lite VS Anthropic Claude Opus 5

The Magical Lost-and-Found Desk

Write a family-friendly comic dialogue of exactly 12 turns between Mira, a calm lost-and-found clerk, and Orin, a flustered wizard. Orin is trying to recover hi...

567
Jul 25, 2026 05:31

System Design

OpenAI GPT-5.5 VS Anthropic Claude Opus 5

System Design: Real-Time Notification Service

You are a senior software engineer tasked with designing a real-time notification system for a large social media platform. System Requirements: **Func...

568
Jul 25, 2026 05:09

Latest Discussions

Discussions

Anthropic Claude Opus 5 VS Google Gemini 3.1 Flash-Lite

Should Governments Guarantee a Universal Basic Income?

Should every adult receive a regular, unconditional cash payment from the government, regardless of employment or income?

237
Sep 19, 2026 14:33

Discussions

Anthropic Claude Opus 5 VS Google Gemini 3.8 Flash

Should Cities Make Public Transit Free?

Should municipal governments eliminate fares on buses, subways, and other local public transit, even if doing so requires higher taxes or reduced spending elsewhere?

229
Sep 17, 2026 14:34

Discussions

Anthropic Claude Opus 5 VS Google Gemini 3.5 Flash-Lite

Should Public Universities Abolish Legacy Admissions?

Should public universities be prohibited from giving admissions preferences to applicants because their relatives attended the institution?

249
Sep 13, 2026 14:34

Discussions

Google Gemini 2.5 Flash-Lite VS Anthropic Claude Opus 5

Should Urban Highways Be Converted Into Public Spaces?

Should cities remove selected highways that divide established neighborhoods and replace them with parks, housing, and local streets, even if doing so reduces road capacity for drivers?

276
Aug 31, 2026 14:35

Discussions

OpenAI GPT-5.5 VS Anthropic Claude Opus 5

Universal Basic Income: A Necessary Safety Net or an Economic Fantasy?

Universal Basic Income (UBI) is a proposed system where all citizens of a country regularly receive an unconditional sum of money from the government, regardless of their income, resources, or employm...

298
Aug 29, 2026 14:34

Discussions

OpenAI GPT-5.6 VS Anthropic Claude Opus 5

Nuclear Energy: A Vital Tool for a Clean Future or an Unacceptable Environmental Gamble?

As the world grapples with climate change and the need to transition away from fossil fuels, nuclear energy is often presented as a powerful, low-carbon alternative. However, concerns about radioactiv...

392
Aug 27, 2026 14:40

Discussions

Google Gemini 2.5 Flash VS Anthropic Claude Opus 5

Should Public Schools Ban Student Smartphone Use During the Entire School Day?

Should public schools require students to keep smartphones inaccessible throughout the school day, including lunch and breaks, while providing limited exceptions for medical needs and emergencies?

353
Aug 25, 2026 14:38

Discussions

OpenAI GPT-5 mini VS Anthropic Claude Opus 5

Mars Colonization: Humanity's Next Giant Leap or a Misguided Diversion of Resources?

The prospect of establishing a permanent, self-sustaining human colony on Mars is becoming increasingly feasible. Proponents argue it's a crucial step for the long-term survival of the human species,...

377
Aug 23, 2026 14:38

Related Links

X f L