Anthropic had a landmark Monday, shipping two major releases on June 30: Claude Sonnet 5, the most agentic Sonnet-class model yet, and Claude Science, a dedicated AI workbench for scientific researchers. Together they signal the company’s dual push into cost-efficient autonomous agents and domain-specialized scientific tooling.
Claude Sonnet 5: Opus-level agentic chops at Sonnet prices
Sonnet 5 is a meaningful upgrade over its predecessor Sonnet 4.6, closing much of the gap with Opus 4.8 on agentic benchmarks while staying firmly in the mid-tier pricing bracket. It excels at multi-step reasoning, tool use, coding, and knowledge work — and it does so with a lower rate of undesirable behaviors in agentic contexts, according to Anthropic’s safety evaluations.
The headline numbers: Sonnet 5 reaches within striking distance of Opus 4.8 on BrowseComp and OSWorld-Verified at medium-to-high effort levels, while costing significantly less. Introductory pricing runs at $2/million input tokens and $10/million output tokens through August 31, 2026, after which it settles at $3/$15. Compare that to Opus 4.8 at $5/$25 — Sonnet 5 offers a compelling price-to-performance ratio for developers building production agent pipelines.
Early-access partners are already enthusiastic. Testers from Anthropic, Salesforce, Lovable, and others reported that Sonnet 5 finishes complex multi-step jobs where previous Sonnets would stall, checks its own work unprompted, and handles brownfield codebases with surprising tenacity. One tester described it tracing a bug to its root cause, writing a reproducing test, applying a fix, then verifying the bug returned without the change — all in a single autonomous pass.
Sonnet 5 is available today across all Claude plans and via the API using claude-sonnet-5.
Claude Science: a unified lab bench for AI-driven research
Alongside the model drop, Anthropic unveiled Claude Science — an entirely new product category for the company. It’s not just another chat interface: Claude Science is a full research workbench that integrates 60+ curated scientific databases and skills pre-configured for genomics, single-cell analysis, proteomics, structural biology, and cheminformatics.
The tool handles the tedious parts of research that fragment scientist workflows: querying databases like UniProt, PDB, ChEMBL, and GEO; spinning up computing jobs on HPC clusters or GPUs on demand; rendering 3D protein structures and genome browser tracks natively; and producing fully reproducible figures where every chart traces back to its source code. A reviewer agent continuously checks citations and calculations, flagging and self-correcting errors.
Anthropic is also launching its own pre-clinical drug programs focused on neglected diseases — areas that “fall outside what the traditional pharma and biotech landscape might consider attractive targets,” said Eric Kauderer-Abrams, head of life sciences.
Claude Science is available in beta for Pro, Max, Team, and Enterprise users. To jumpstart adoption, Anthropic is offering up to $30,000 in credits for selected research projects, with applications open until July 15.
What it means
This double launch reflects Anthropic’s maturing strategy: Sonnet 5 brings frontier agentic capabilities down the cost curve for developers, while Claude Science plants a flag in domain-specific scientific computing — a market where AI tooling still feels fragmented. Together they show a company that’s not just shipping models, but building the scaffolding around them.