AnthropicAnthropicΒ·text

Claude Opus 4.8

CodeVisionReasoningWeb searchFunction calling
Advertised as anthropic/claude-opus-4.8claude-opus-4-8claude-opus-4.8Claude-Opus-4.8opus-4-8opus-4.8
Quick reference
Claude Opus 4.8 β€” TLDR
  • 🏒 Anthropic's most capable generally available Opus model, released May 2026.
  • πŸ“ 1M-token context window by default; 128K max output tokens.
  • 🧠 Adaptive thinking, defaulting to high effort across all surfaces.
  • πŸ”§ Better tool triggering and fewer agentic-coding compactions than Opus 4.7.
  • πŸ‘οΈ Multimodal: accepts text, image, and file inputs.
  • 🎯 Roughly four times less likely than predecessor to leave code flaws unremarked.
  • ⚑ Optional fast mode (research preview) reduces latency and cost.
  • 🌐 Web search, function calling, and memory-driven long-horizon work.
πŸ’° Best price on Antseed
$0.094 / $0.284βˆ’98%
per 1M Β· cheapest in / out
🐜 Providers
41
advertising on Antseed
πŸ”¬ Model checks
Not audited yet
no audit of this model so far
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 4.8 is Anthropic's most capable generally available model, launched on May 28, 2026, and positioned for complex reasoning, long-horizon agentic coding, and high-autonomy knowledge work. It supports a 1M-token context window by default on the Claude API, Amazon Bedrock, and Vertex AI (200K on Microsoft Foundry), up to 128K output tokens, and adaptive thinking that defaults to high effort. It accepts text, image, and file inputs with text output.

The model is a direct upgrade to Claude Opus 4.7, itself following Claude Opus 4.6 and Claude Opus 4.5. Anthropic describes it as a "modest but tangible improvement," with gains concentrated in long-horizon agentic codingβ€”better long-context handling, fewer compactions, and better compaction recoveryβ€”plus improved tool triggering, addressing an issue some users reported on Opus 4.7. Anthropic reports the model is about four times less likely than Opus 4.7 to leave flaws in its own code unremarked, reflecting a focus on honesty and coherence over many turns.

On developer experience, the minimum cacheable prompt length drops to 1,024 tokens, and developers can now update instructions mid-task without breaking the prompt cache. These changes target the iterative, tool-heavy workflows the model is built for.

This release also introduces a fast-mode research preview, mirrored by the separate Claude Opus 4.8 Fast variant, alongside siblings including Claude Sonnet 4.6 and the newer Claude Fable 5.

Sources
platform.claude.comModels overview - Claude API DocsΒ· platform.claude.comanthropic.comClaude Opus 4.6 \ 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 4.8
Tokens served
493.34M
input + output
Requests
17,651
settled calls
Buyers
93
distinct, on this model
Providers used
35
of 41 advertising
Settled
$626.19
gross USDC, this model

Daily usageΒ Β·Β Claude Opus 4.8Β Β·Β Organic providers

Top providersΒ Β·Β Claude Opus 4.8Β Β·Β Organic

Providers serving Claude Opus 4.8 (41)compare on the network explorer 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.