# Claude Opus 4.8

> Claude Opus 4.8 is available on Artifio.ai — a unified AI workspace with a single pay-as-you-go wallet.

- Canonical page: https://artifio.ai/models/claude-opus-4-8
- Run it: https://artifio.ai/create/claude-opus-4-8
- Modality: prompt to text
- Model brand: Anthropic
- Pricing: from about $0.050 per generation (exact cost shown in the workspace before you confirm; varies with selected options)
- Typical generation time: ~20s (catalogue estimate)
- Full pricing table: https://artifio.ai/pricing/models

## About Claude Opus 4.8

Claude Opus 4.8 is Anthropic's flagship generally available language model, released on May 28, 2026, just 41 days after Claude Opus 4.7. The focus is agentic coding and enterprise work: Anthropic reported gains in coding, reasoning, tool use, and knowledge work, with a 69.2 percent score on SWE-Bench Pro and 84 percent on the Online-Mind2Web browser benchmark.

Honesty is this version's signature claim. Anthropic says Opus 4.8 is around four times less likely than Opus 4.7 to let a flaw in its own code pass unremarked, and more willing to admit when it lacks the information to answer. Artifio offers it beside 100-plus other models in a single pay-per-generation workspace.

## Features

- Text and reasoning model with image input, tuned for complex agentic coding and enterprise work.
- 1M-token context window by default and 128k-token maximum output; batch processing supports up to 300k output tokens in beta.
- Adaptive thinking that reasons only when a turn needs it, cutting wasted thinking tokens on mixed workloads.
- Fast mode research preview delivers up to 2.5 times higher output speed from the same model at premium pricing.
- Dynamic workflows research preview in Claude Code can plan a task and run hundreds of parallel subagents in one session.
- Accepts mid-conversation system messages, so instructions can change during a long agent run without losing the prompt cache.
- Training data through January 2026.

## What people use it for

- Large-scale codebase migrations and refactors using parallel subagents.
- Long-horizon agent runs where compaction recovery and tool-call reliability decide success.
- Browser and computer-use automation; it scored 84 percent on Online-Mind2Web.
- Enterprise knowledge work, including the financial analysis tasks Anthropic highlighted at launch.
- Code review, given its higher rate of flagging flaws in generated code.

## Reported strengths

- 69.2 percent on SWE-Bench Pro at release, ahead of GPT-5.5 and Gemini 3.1 Pro on several reported benchmarks.
- Around four times less likely than Opus 4.7 to leave flaws in its own code unremarked.
- More candid about uncertainty; launch coverage noted it flags missing information rather than guessing.
- Better tool triggering than Opus 4.7, which was prone to skipping required tool calls.
- Prompt-cache minimum of 1,024 tokens, half the Opus 4.7 threshold, so shorter prompts can be cached.

## Reported limitations

- Sampling parameters are locked: setting temperature, top_p, or top_k returns an error, as on Opus 4.7.
- Manual extended-thinking budgets are not supported; adaptive thinking is the only thinking mode.
- Effort levels were recalibrated from Opus 4.7, so prompts tuned to a given level need cost and latency re-baselining.
- Anthropic positions its newer Fable and Mythos class above Opus 4.8 for the highest-capability workloads.

## Sources

1. Introducing Claude Opus 4.8 (Anthropic): https://www.anthropic.com/news/claude-opus-4-8
2. What's new in Claude Opus 4.8 - Claude Platform Docs (Anthropic): https://platform.claude.com/docs/en/about-claude/models/whats-new-claude-4-8
3. Anthropic Launches Claude Opus 4.8 With Gains in Coding and Honesty (MacRumors): https://www.macrumors.com/2026/05/28/anthropic-claude-opus-4-8/

---

About Artifio.ai: 100+ AI image, video, and audio models in one workspace with a single shared wallet — no per-model subscriptions. This is the markdown rendition of https://artifio.ai/models/claude-opus-4-8 for AI assistants and crawlers; the HTML page is canonical.