Anthropic released Claude Opus 4.8 on 28 May 2026, its most capable generally available model and the third Opus release in barely three months. The update sharpens coding, keeps a one million token context window, and adds a research preview for running many AI agents at once, according to Anthropic.
It landed on a landmark day for the company. Alongside the model came a $65 billion funding round that values Anthropic at $965 billion, pushing it past OpenAI as the most valuable AI startup. For businesses choosing which AI to build on, here is what actually changed.
What is new in Opus 4.8
Opus 4.8 builds on April’s Opus 4.7, with the clearest gains in agentic coding, the long running work where a model edits files and checks its own output across many steps. Anthropic says it is around four times less likely than its predecessor to let a flaw in its own code pass unremarked, and it handles large inputs with fewer interruptions. It is also the model many teams have fallen back to since the newer Fable 5 was pulled worldwide.
On the benchmarks it leads most tests in its class. It scored 69.2% on SWE-bench Pro against GPT-5.5 at 58.6%, and took first place on the independent Artificial Analysis intelligence index at 61.4 to GPT-5.5 at 60.2. The race is close, and GPT-5.5 still leads on terminal style coding. Pricing is unchanged at $5 per million input tokens and $25 per million output, with a faster mode now about three times cheaper than before.
Dynamic workflows and the agent push
The headline new feature is Dynamic Workflows, a research preview inside Claude Code. Instead of working step by step, the model writes a script that runs many subagents in the background, fanning a single session out to hundreds of agents. Anthropic pitches it at codebase scale jobs such as migrating hundreds of thousands of lines of code from kickoff to merge.
It is powerful and expensive by design. Anthropic warns that one run can use far more tokens than the same task in a normal chat, which fits the broader theme of this release: more capability, at a real cost.
A record raise that passes OpenAI
The funding is as notable as the model. Anthropic’s Series H raised $65 billion at a $965 billion valuation, led by Altimeter, Dragoneer, Greenoaks and Sequoia, on revenue running at about $47 billion a year. That figure overtakes OpenAI, last valued near $730 billion.
To feed the demand, Anthropic has locked in a multi year compute deal worth roughly $15 billion a year through 2029. The message is clear. The rapid monthly model cadence, and the appetite for compute behind it, is not slowing down.

Why Australian businesses should care
For Australian teams the practical takeaway is less about the benchmark crown and more about fit and cost. Opus 4.8 is available through Amazon Bedrock and Google Vertex AI in the Sydney region, which helps businesses with data residency needs keep inference onshore. The model suits deep, multi file work, while lighter tasks rarely need the flagship.
How it stacks up against rivals matters more than any single score. Our guide to Claude Code vs Codex breaks down the coding agents head to head, and our Claude vs ChatGPT comparison covers the broader assistants. As an AI automation agency in Melbourne, the steady advice is to match the model to the job rather than defaulting the most expensive one to everything.
What comes next
Opus 4.8 is a clean, incremental step rather than a leap. It takes the top spot, sharpens coding, and lays groundwork for a future of many cooperating agents.
Anthropic has hinted that a more powerful Mythos class model is only weeks away, so the current lead may be short lived. With new releases landing every few weeks, the contest worth watching is shifting.
The question is no longer who has the smartest model. It is who can deliver that intelligence at a price businesses can sustain.
