News
Research Report
The AI Transformation 100: Concrete Ideas for Fixing How We Work
Key Takeaways
- Anthropic released Claude Opus 4.8 with improved coding, reasoning and agentic task performance.
- Claude Code now supports dynamic workflows that can run hundreds of parallel subagents on large software tasks.
- Opus 4.8 keeps standard pricing unchanged while adding effort controls and cheaper fast mode pricing.
Anthropic released Claude Opus 4.8 with improved coding, reasoning and agentic task performance.Claude Code now supports dynamic workflows that can run hundreds of parallel subagents on large software tasks.
Opus 4.8 keeps standard pricing unchanged while adding effort controls and cheaper fast mode pricing.
Anthropic released Claude Opus 4.8 on Wednesday, upgrading its flagship model with improvements in coding, agentic task performance, reasoning and practical knowledge work.
"It's our strongest coding model yet," Boris Cherny, head of Claude Code, wrote on X. "...up on SWE-bench Pro (from 64.3 to 69.2) and noticeably more honest about its own work. It tells you when it's unsure and catches its own bugs instead of declaring victory early. Same price as 4.7."
The new model is available immediately at the same regular pricing as Claude Opus 4.7. Developers can access it through the Claude API using claude-opus-4-8, while users can access the model across Anthropic’s products.
The release comes with several product updates, including effort controls in claude.ai, a new dynamic workflows feature for Claude Code and lower pricing for Opus 4.8’s fast mode. Anthropic said fast mode allows the model to work at 2.5 times the speed and is now three times cheaper than it was for previous models.
Table of Contents
- Claude Opus 4.8 Adds Coding and Agentic AI Improvements
- Anthropic Says Opus 4.8 Shows Lower Misalignment Risk
- Claude Code Gets Dynamic Workflows
- Users Can Control How Much Effort Claude Uses
- Anthropic Updates Messages API for Developers
- Opus 4.8 Pricing Remains Unchanged
- Anthropic Teases Mythos-Class Models
Claude Opus 4.8 Adds Coding and Agentic AI Improvements
Anthropic described Opus 4.8 as a “modest but tangible improvement” over Opus 4.7, with gains across benchmarks and stronger performance as a collaborator on complex tasks. The company said early testers found the model more reliable and sharper in its judgment when handling agentic work.
One of the more notable changes, according to Anthropic, is improved honesty. The company said Opus 4.8 is less likely to make unsupported claims about progress, a common issue with AI models that can confidently report success even when evidence is thin.
Anthropic said its evaluations found Opus 4.8 was about four times less likely than Opus 4.7 to let flaws in code it had written pass without comment, adding that the improvement is part of a broader effort to make AI systems better at flagging uncertainty and avoiding overconfident answers.
Anthropic Says Opus 4.8 Shows Lower Misalignment Risk
The model also underwent an alignment assessment before release. Anthropic said its Alignment team found Opus 4.8 reached new highs on measures tied to prosocial behavior, including supporting user autonomy and acting in the user’s best interest.
The company said the model also showed substantially lower rates of misaligned behavior than Opus 4.7, including deception or cooperation with misuse. Those results put Opus 4.8 closer to Claude Mythos Preview, which it described as one of its best-aligned models.
Claude Code Gets Dynamic Workflows
Alongside the model release, Anthropic introduced dynamic workflows for Claude Code in research preview. The feature is designed to let Claude take on larger software engineering tasks by planning work, running hundreds of parallel subagents in a single session and verifying outputs before reporting back to the user.
Anthropic said the feature can support codebase-scale migrations across hundreds of thousands of lines of code, using an existing test suite as the standard for completion. Dynamic workflows are available in Claude Code for Enterprise, Team and Max plans.
Users Can Control How Much Effort Claude Uses
The company also added effort control in claude.ai and Cowork. The new setting appears alongside the model selector and lets users choose how much effort Claude applies to a response.
Higher effort settings prompt the model to think more frequently and deeply, while lower settings prioritize faster answers and slower rate-limit usage. Anthropic said effort control is available on all plans.
Opus 4.8 defaults to high effort, which Anthropic said offers the best overall balance between quality and user experience. For coding tasks, the company said high effort uses a similar number of tokens as Opus 4.7’s default setting while delivering better performance.
Users can also choose higher settings, including “extra” and “max,” for more difficult tasks and longer-running workflows.
Anthropic Updates Messages API for Developers
Anthropic also updated its Messages API to accept system entries inside the messages array. The change allows developers to update Claude’s instructions during a task without breaking the prompt cache or routing the update through a user turn.
The change can be used to adjust permissions, token budgets or environment context while an agent is running.
Opus 4.8 Pricing Remains Unchanged
Pricing for standard Opus 4.8 usage remains unchanged from Opus 4.7 at $5 per million input tokens and $25 per million output tokens.
Fast mode pricing is set at $10 per million input tokens and $50 per million output tokens. According to Anthropic, fast mode lets Opus 4.8 work at 2.5 times the speed and is now three times cheaper than it was for previous models.
Anthropic Teases Mythos-Class Models
Anthropic also noted more model releases ahead, adding that it is working on models that provide many Opus-level capabilities at lower cost, as well as a new class of model with higher intelligence than Opus.
Webinar
Jun
3
From AI Investment to CX Results: What Enterprise Leaders Need to Know
Move beyond experiments. See how top enterprises scale AI for CX results.
Register
Webinar
Jun
4
Content Strategy Leaders Live: Managing Risk, Compliance & AI in Financial Services
Learn how financial services leaders are modernizing content systems without disrupting trust, compliance or experience.
Register
Webinar
Jun
16
From Legacy to Launch-Ready: How Gainbridge Made Its Website a Marketing-Led Growth Engine
Join in to learn how a D2C annuity brand gave marketing full website ownership — without slowing down or risking compliance.
Register
Webinar
On demand
Content Strategy Leaders Live: Managing Scale, Safety & AI in Manufacturing
How manufacturing leaders update content systems to handle product complexity and scale while integrating AI safely.
Watch Now
Webinar
On demand
AI With Intent: How CX Leaders Decide Where Automation Belongs (And Where It Doesn't)
Senior CX leaders from REI & Hyatt reveal their framework for using AI to boost retention.
Watch Now
Webinar
On demand
From Content Sprawl to Competitive Advantage: A Knowledge Operations Roadmap
How automation, AI and a centralized content platform turn knowledge fragmentation into a scalable competitive advantage.
Watch Now
That work is tied to Project Glasswing, under which a small number of organizations are currently using Claude Mythos Preview for cybersecurity work. Anthropic said models at that capability level require stronger cyber safeguards before broader release, but the company expects to bring Mythos-class models to customers in the coming weeks.