Anthropic releases Claude Opus 4.8 with 69.2% agentic coding score, 2.5x faster performance
Anthropic released Claude Opus 4.8 on May 28, 2026, six weeks after version 4.7. The model achieves 69.2% on agentic coding benchmarks (up from 64.3%), runs 2.5 times faster in fast mode at one-third the cost, while maintaining the same pricing as version 4.7.
Claude Opus 4.8 — Quick Specs
Anthropic releases Claude Opus 4.8 with 69.2% agentic coding score, 2.5x faster performance
Anthropic released Claude Opus 4.8 on May 28, 2026, just six weeks after the previous version 4.7 launched on April 16. The release accelerates Anthropic's model update cadence while maintaining pricing parity with the previous version.
Benchmark improvements
According to Anthropic, Claude Opus 4.8 shows measurable gains across five benchmark categories:
- Agentic coding: 69.2% (up from 64.3% in version 4.7)
- Multidisciplinary reasoning with tools: 57.9% (up from 54.7%)
- Agentic computer use: 83.4% (up from 82.8%)
- Knowledge work: 1890 (up from 1753)
- Agentic financial analysis: 53.9% (up from 51.5%)
The model's fast mode now runs approximately 2.5 times faster than version 4.7 while costing three times less to operate, though Anthropic did not disclose absolute pricing figures.
New capabilities and features
Opus 4.8 introduces what Anthropic calls "high effort" mode as the default for coding tasks, which the company claims uses a similar token budget to Opus 4.7 while delivering better performance. Users can select "extra" (labeled "xhigh" in Claude Code) or "max" settings to allocate more tokens for improved results on difficult tasks.
Anthropic launched three additional features alongside the model release:
- Dynamic workflows: A research preview feature in Claude Code that allows the model to handle larger-scale tasks
- Effort control: Available in claude.ai and Cowork, letting users adjust computational effort per response
- Messages API update: Developers can now insert system entries within the messages array to update instructions mid-task without breaking the prompt cache
According to Anthropic, early testers report that Opus 4.8 more frequently flags uncertainties and makes fewer unsupported claims compared to previous versions. The company describes the model as having "sharper judgement, more honesty about its progress, and the ability to work independently for longer."
Mythos model coming soon
Anthropic stated it will release its Mythos-class cybersecurity model to all customers "in the coming weeks." The company unveiled Mythos in early April 2026 but has limited access to select stakeholders managing key software platforms.
Claud Opus 4.8 is available globally as of May 28, 2026. Pricing remains unchanged from version 4.7, though specific per-token costs were not disclosed.
What this means
The six-week release cycle represents a significant acceleration in Anthropic's development timeline. The 4.9 percentage point improvement in agentic coding (64.3% to 69.2%) suggests meaningful progress in code generation and autonomous task execution, while the 2.5x speed increase in fast mode addresses a common deployment constraint. The decision to maintain pricing while improving performance indicates competitive pressure in the AI model market, particularly from recent releases by OpenAI and other providers.
Related Articles
Anthropic Paper: Automated AI Researchers Beat Humans at Alignment Fixes for $4/Hour
A new Anthropic paper from its fellows program shows an automated AI system improving performance on all 10 tested alignment benchmarks, outperforming experienced human researchers within six hours at a fraction of the cost. The research, led by Anthropic Fellow Chen Yueh-Han, is described as early evidence that automated alignment post-training could become practical soon.
Tencent Open-Sources Hy4 Preview: 770B-Parameter MoE Model with 1M-Token Context
Tencent's Hy Team has open-sourced Hy4 preview, a 770-billion-parameter Mixture-of-Experts model with 49 billion activated parameters and a 1-million-token context window. The model is available under Apache 2.0 alongside an FP8-quantized variant, with Tencent claiming it beats GLM 5.3 and Kimi K3 on internal engineering evaluations.
Tencent Releases Hy4 Preview: 770B-Parameter MoE Model with 1M Context for Coding Agents
Tencent has released Hy4 preview, a mixture-of-experts model with 770B total parameters and 49B active parameters, targeting coding agents and multi-step tool-use workflows. The model ships with a 1 million token context window and is priced at $0.834 per 1M input tokens and $2.501 per 1M output tokens.
Anthropic Adds Built-In Browser to Claude Cowork Desktop App
Anthropic is embedding a dedicated browser into Claude Cowork's desktop app, opening in a side panel whenever a task requires web access. The browser is isolated from the user's own tabs, bookmarks, and passwords, and rolls out this week to Pro, Max, Team, and Enterprise plans.
Comments
Loading...