Anthropic

Claude Opus 5.5

Anthropic's new Opus. Ahead of Fable 5.1 on every published test, at 20% less per token than Opus 5.

successor to claude-opus-5text + image inputtext output

Replaces Claude Opus 5: input $5.25 → $4.2 per 1M (-20%), output $26.25 → $21 per 1M (-20%), same context window.

Run this model

anthropic/claude-opus-5.5

Use an endpoint supported by this model.

Get a key for this setup →
const response = await fetch('https://api.minirouter.sh/v1/chat/completions', {  method: 'POST',  headers: {    Authorization: `Bearer ${process.env.MINIROUTER_KEY}`,    'Content-Type': 'application/json',  },  body: JSON.stringify({    model: 'anthropic/claude-opus-5.5',    messages: [{ role: 'user', content: 'Why is the sky blue?' }],    max_tokens: 1024,  }),}) const data = await response.json()console.log(data.choices[0].message.content)
Read docs →

Model overview

Input / 1M
$4.2
All-in MiniRouter rate
Output / 1M
$21
All-in MiniRouter rate
Context
1M
1,000,000 tokens
Max output
128K
tokens
Terminal-Bench 4.0
66.4%
Opus 5 scored 52.3%
Cache read
$0.20 list
was $0.50 on Opus 5
Announced
22 Sept 2026
publisher release

Publisher-reported, plotted by MiniRouter

A clear step past Opus 5.

Opus 5.5 leads the Claude field on all eight. GPT-6 Astra is ahead on AutomationBench and Terminal-Bench-Science.

Anthropic-reported

Coding, agents, knowledge work

  • Claude Opus 5.5
  • Comparison models

Agentic coding

Terminal-Bench 4.0

  • Claude Opus 5.566.4%
  • Claude Fable 5.155.8%
  • Claude Opus 552.3%
  • GPT-6 Astra57.9%

0 to 100%

Coding

FrontierCode v1.1

  • Claude Opus 5.554.4%
  • Claude Fable 5.150.3%
  • Claude Opus 548.0%
  • GPT-6 Astra53.3%

0 to 100%

Agentic coding

CursorBench 4.0

  • Claude Opus 5.557.8%
  • Claude Fable 5.151.8%
  • Claude Opus 546.6%
  • GPT-6 Astra—Not reported

0 to 100%

Knowledge work

GDPval-AA v2.1 (Elo)

  • Claude Opus 5.51,846
  • Claude Fable 5.11,735
  • Claude Opus 51,708
  • GPT-6 Astra1,542

1,500 to 1,900 Elo

Business workflows

AutomationBench

  • Claude Opus 5.540.0%
  • Claude Fable 5.131.4%
  • Claude Opus 526.9%
  • GPT-6 Astra41.4%

0 to 100%

Reasoning

Humanity's Last Exam · with tools

  • Claude Opus 5.567.7%
  • Claude Fable 5.165.6%
  • Claude Opus 563.6%
  • GPT-6 Astra57.2%

0 to 100%

Agentic research

Terminal-Bench-Science 0.1

  • Claude Opus 5.558.7%
  • Claude Fable 5.152.6%
  • Claude Opus 529.0%
  • GPT-6 Astra64.6%

0 to 100%

Computer use

OSWorld 2.0 · partial

  • Claude Opus 5.581.8%
  • Claude Fable 5.180.7%
  • Claude Opus 574.0%
  • GPT-6 Astra—Not reported

0 to 100%

Max effort, production safeguards on. Terminal-Bench 4.0 at xhigh effort. Source: Anthropic announcement ↗, September 22, 2026.
View chart data
Values from Anthropic's Claude Opus 5.5 launch table
BenchmarkClaude Opus 5.5Claude Fable 5.1Claude Opus 5GPT-6 Astra
Terminal-Bench 4.066.4%55.8%52.3%57.9%
FrontierCode v1.154.4%50.3%48.0%53.3%
CursorBench 4.057.8%51.8%46.6%Not reported
GDPval-AA v2.1 (Elo)1,8461,7351,7081,542
AutomationBench40.0%31.4%26.9%41.4%
Humanity's Last Exam · with tools67.7%65.6%63.6%57.2%
Terminal-Bench-Science 0.158.7%52.6%29.0%64.6%
OSWorld 2.0 · partial81.8%80.7%74.0%Not reported

Pricing

Cheaper per token, fewer tokens per task.

Anthropic puts typical workloads at about 40% cheaper than Opus 5.

Opus 5.5 Fast

Anthropic list, per 1M tokens

Input
$4.00was $5.00
Output
$20.00was $25.00
Cache write
$5.00
Cache read
$0.20was $0.50
Source: Anthropic announcement ↗, September 22, 2026.
View chart data
List prices per 1M tokens
RateOpus 5.5Opus 5.5 FastOpus 5
Input$4.00$8.00$5.00
Output$20.00$40.00$25.00
Cache read$0.20$0.40$0.50
Cache write$5.00$10.00$6.25

One route, standard clients.

Authenticate with a MiniRouter bearer key. Caller-provided upstream credentials are not used on this route.

Read model setup docs →
Base URLhttps://api.minirouter.sh/v1
Model IDanthropic/claude-opus-5.5
Endpoints
/v1/chat/completions/v1/messages/v1/responses
Parameters

max_tokens, stop, tools, tool_choice, reasoning, include_reasoning

Structured output

Not enforced; a requested format is dropped and the answer is plain text.

Configured provider

Priority follows the live catalog order.

Input: text, image · Output: text

ProviderPriorityInput / 1MOutput / 1MParametersAccess
vercel
1$4.2$21max_tokens, stop, tools, tool_choice, reasoning, include_reasoningMiniRouter balance