AI Agent Hub
Back to models
Anthropic: Claude Opus 5.5 (batch) logo

Anthropic: Claude Opus 5.5 (batch)

Closed Source anthropic Released 2026-09-22
58.0 / 100 1M context Proprietary

About this model

Claude Opus 5.5 (batch) is Anthropic's flagship Opus-tier model exposed through the Batch API, which applies the same claude-opus-5-5 weights and capabilities as the synchronous API while offering roughly 50% lower input and output pricing for workloads that tolerate asynchronous completion. It is positioned for long-running agentic coding, computer use, and professional knowledge work, with always-on adaptive extended thinking, up to a 1M-token context window, and up to 300K tokens of output per request in batch (beta). Inputs may include text and images; outputs are text.

Relative to earlier Opus generations, Anthropic reports broad capability gains across agentic software engineering, terminal environments, multimodal reasoning, and economically valuable professional tasks, with strong scores on SWE-bench variants, DeepSWE, Terminal-Bench, Humanity's Last Exam, GDPval-AA, and independent agent benchmarks. The batch endpoint suits offline evaluation pipelines, large-scale data generation, and backlog processing where latency is less critical than unit cost.

The model is proprietary and served only via Anthropic's cloud APIs and major cloud partners (Amazon Bedrock, Google Cloud Vertex, Microsoft Foundry, Claude Platform on AWS). Parameter count and training architecture are not published; reliable knowledge is current through June 2026. Standard API list pricing is $4 per million input tokens and $20 per million output tokens, with batch billed at half those rates before caching and tool surcharges.

Benchmark Scores

HLE
63.6
MMLU
94.3
DeepSWE
73.65
MMLU-Pro
93.4
MMMU-Pro
84.7
GDPval-AA
1846.0
LiveBench
80.08
BrowseComp
90.8
SimpleBench
80.6
GPQA-Diamond
93.7
HLE-Verified
54.4
Context-Arena
97.72
NL2Repo-Bench
75.3
SWE-Bench-Pro
79.2
AutomationBench
50.3
Agents-Last-Exam
32.2
Creative-Writing
2132.6
SWE-Bench-Verified
96.0
Terminal-Bench-2.1
89.1
Terminal-Bench-3.0
43.3
Toolathlon-Verified
77.8
SWE-Bench-Multilingual
89.5

Technical Specs

  • Architecture: Transformer
  • Context Window: 1,000,000 tokens
  • Input Modalities: text, image

Hardware Requirements

  • Compute: API only

Pricing

Input Output Currency
2.00 / 1M tokens 10.00 / 1M tokens USD