AnthropicAnthropicΒ·text

Claude Opus 5.5

CodeVisionReasoningWeb searchFunction calling
Advertised as anthropic/claude-opus-5.5claude-opus-5-5Claude-Opus-5-5claude-opus-5.5opus-5-5opus-5.5
Quick reference
# Claude Opus 5.5 β€” TLDR
  • - 🏒 Anthropic's flagship Opus-tier model, succeeding Claude Opus 5.
  • - πŸ“ 1M-token context window with 128K max output tokens.
  • - 🧠 Adaptive thinking always on; effort is the main control lever.
  • - πŸ”§ Built for long-running agentic coding, tool loops and computer use.
  • - ⚑ Anthropic reports ~40% lower running cost than Opus 5 on typical workloads.
  • - πŸ‘οΈ Multimodal inputs, function calling and web search supported.
  • - πŸ†• Breaking changes: thinking can't be disabled, forced tool use errors.
  • - πŸ“š Knowledge work gains: documents, analysis, long-running collaboration.
πŸ’° Best price on Antseed
$0.050 / $0.250βˆ’99%
per 1M Β· cheapest in / out
🐜 Providers
31
advertising on Antseed
πŸ”¬ Model checks
16 of 21 providers pass
5 failed Β· 18 Nov Β· details β†’
πŸ“ Context
1M tokens
Provider

Anthropic PBC is an American artificial intelligence company headquartered in San Francisco. Structured as a public benefit corporation, the lab develops large language models under the Claude name, with a research emphasis on building reliable, steerable, and safety-focused AI…

Explore 18 more models by Anthropic β†’
About this model

Claude Opus 5.5 is Anthropic's top-tier Claude model, released in 2026 as the direct successor to Claude Opus 5, which itself was a step change over Claude Opus 4.8. Anthropic positions it for long-running agentic coding and knowledge work, with a one-million-token context window, up to 128K output tokens per response (and, on the batch API with a beta header, up to 300K), plus vision input, tool calling and web search.

The clearest generational change is efficiency. Anthropic states that Opus 5.5 costs about 40% less to run than Opus 5 on typical workloads, and that on Terminal-Bench 4.0 β€” a multi-step command-line task benchmark β€” Opus 5.5 at default effort surpasses Opus 5 at maximum effort for roughly a fifth of the cost, per the company's own reported results.

Behaviourally, thinking is now always adaptive and cannot be switched off, making the effort setting the primary lever for trading depth against latency and cost. Migrating from Opus 5 involves four breaking changes: thinking cannot be disabled, forced tool use returns an error, thinking blocks are bound to a specific model and conversation, and the older computer-use tool version is no longer accepted on the Claude API and Google Cloud.

Within the wider lineup it sits above the Sonnet tier, including Claude Sonnet 5, alongside Claude Fable 5.1 and the earlier low-latency Claude Opus 5 Fast.

Sources
platform.claude.comClaude Opus 5 - Claude Platform DocsΒ· platform.claude.comanthropic.comIntroducing Claude Opus 5 \ AnthropicΒ· anthropic.com

This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.

Usage on AntseedΒ Β·Β Claude Opus 5.5
Tokens served
27.60B
input + output
Requests
571,849
settled calls
Buyers
306
distinct, on this model
Providers used
44
of 31 advertising
Settled
$11.69k
gross USDC, this model

Daily usageΒ Β·Β Claude Opus 5.5Β Β·Β Organic providers

Top providersΒ Β·Β Claude Opus 5.5Β Β·Β Organic

Providers serving Claude Opus 5.5 (31)compare on the network explorer β†’model checks β†’
ProviderHealthRep↓SalesTokensRoutingInput $/MCached
$/M
Output $/MSave1st tokentok/s

"Best price" and the provider table are live Antseed catalog data (advertised $/1M tokens β€” or $ per generated image for unit-billed image models β€” not settled amounts). Reputation = buyer trust score (0-100, the Antseed SDK's own formula). "Routing" = the SDK's default buyer routing (what Antseed Desktop ships with): a trust β‰₯ 60 gate on the effective reputation, then cheapest-first among routable providers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on Antseed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.