Updated

A weathered terracotta clay vessel with ivory paper strips wrapped around it on a dark background, abstract hero artwork for Claude Opus 4.8

Anthropic

Claude Opus 4.8

Anthropic's Opus 4.8, about four times less likely than Opus 4.7 to let flaws in its code pass unremarked.

Chat with Claude Opus 4.8 in Super AgentView pricing

What is Claude Opus 4.8?

Claude Opus 4.8 is the Opus model Anthropic released on May 28, 2026.

Anthropic described it as a modest but real improvement over Opus 4.7 across benchmarks and a more effective collaborator at the same price. It is also more honest: about four times less likely than its predecessor to let flaws in code it wrote pass without comment, and its rate of misaligned behavior is substantially lower than Opus 4.7's. Anthropic now lists it as a legacy model and recommends Claude Opus 5.5.

Provider
Anthropic
Context window
1M tokens
Input
Text, Image
Reasoning
Off to max
Max output
128K tokens
Knowledge cutoff
January 2026
Released
May 28, 2026

How Claude Opus 4.8 performs

Published by Anthropic

Results Anthropic reported for Claude Opus 4.8. Each lab runs its own setup, so compare these with other models only inside the same table.

Benchmark results Anthropic published for Claude Opus 4.8
BenchmarkScoreSetting
SWE-Bench Pro69.2%
Terminal-Bench 2.174.6%Terminus-2 harness
Humanity's Last Exam49.8%no tools
Humanity's Last Exam57.9%with tools
OSWorld-Verified83.4%
Finance Agent v253.9%

Source: anthropic.com

What you can do with Claude Opus 4.8

  • About four times less likely than Opus 4.7 to stay quiet about flaws in its own code

  • Lower rate of misaligned behavior than Opus 4.7, per Anthropic

  • Thinking can be switched off or raised to max

  • 1M tokens of context with text and image input

How to prompt Claude Opus 4.8

  • Ask it to list the weak spots in its own answer; Anthropic says it is more upfront about flaws.
  • Keep one long conversation for a project so it can build on earlier steps.
  • Super Agent starts it at medium effort. Turn reasoning off for quick edits.

Where Claude Opus 4.8 fits best

  • Code review where you want problems called out

  • Analytical work over many steps

  • Long sessions on one project

  • Comparing Opus generations

How Claude Opus 4.8 compares

Claude Opus 4.8 compared with other chat models on Flashloop
ModelLMArena textLMArena codingContext windowReasoning
Claude Opus 4.81482#251533#181M tokensOff to max
GLM-5.31481#261521#331M tokensLow, medium, high, max
Gemini 3.6 Flash1483#221518#411M tokensLow, medium, high
GPT-6.1 Sol1480#271537#141.05M tokensLow to max
GPT 5.6 Sol1485#211534#171.05M tokensOff to max

Good to know before you start

  • Anthropic lists it as legacy; Opus 5 and Opus 5.5 are newer
  • Anthropic calls it a modest step up from Opus 4.7
  • Its long outputs are text; images and videos in Super Agent come from Flashloop's generation models

Start chatting with Claude Opus 4.8

Chat with Claude Opus 4.8 in Super Agent

  1. Select Claude Opus 4.8

    Open Super Agent on Flashloop and choose Claude Opus 4.8 in the model picker.

  2. Enter your prompt

    Ask for what you need: an answer, a plan, or the images and videos you want the agent to make.

  3. Send and keep going

    Send the message. Claude Opus 4.8 answers in Super Agent and runs any generations you asked for.

Questions about Claude Opus 4.8

What is Claude Opus 4.8 best at?

Anthropic positioned it as a steady step up from Opus 4.7, with better benchmark scores, more honest reporting of flaws in its own code and less misaligned behavior.

Is Claude Opus 4.8 Anthropic's newest Opus?

No. Claude Opus 5 and Claude Opus 5.5 came after it. All three are in Super Agent.

What reasoning levels does Claude Opus 4.8 support?

Off, low, medium, high, extra high and max. Super Agent starts it at medium.

Where is Claude Opus 4.8 in the model picker?

Under All models.

More chat models on Flashloop

Browse all models