Anthropic shipped Claude Sonnet 5 on June 30, and the pitch is blunt: this is now the cheapest way to run agents that don’t fall apart. It’s a mid-tier LLM built for agentic workflows — planning, calling tools, driving a browser or terminal, and grinding through long coding tasks without a human babysitting every step.
Why people are paying attention
The numbers are the story. Sonnet 5 hits 63.2% on SWE-bench Pro (up from 58.1% on Sonnet 4.6), jumps Terminal-Bench 2.1 from 67% to 80.4%, and posts 81.2% on OSWorld computer-use. On the GDPval-AA knowledge-work test it scores 1618 — actually nudging past Opus 4.8’s 1615. So you’re paying mid-tier money for near-flagship output. It’s already the default model on claude.ai, and HN lit up with 711 points the day it dropped.
Using it through the API
Call it with model ID claude-sonnet-5-20260401. Pricing is $3/$15 per million input/output tokens, with an intro rate of $2/$10 through August 31. Opus 4.8 costs $5/$25. For anyone running coding agents or multi-step tool loops at volume, that gap compounds fast — same workload, roughly half the bill.
You Might Also Like
- Kimi k2 6 Beats gpt 5 4 and Claude Opus 4 6 on swe Bench pro
- Z ais glm 5 1 Tops swe Bench pro at 58 4 Trained on Zero Nvidia Hardware
- Mistral Medium 3 5 Scores 77 6 on swe Bench one Point shy of Gemini 3 1 pro
- Hiveterm Bets on the Multi Agent Workspace Claude Codex and Gemini in one Terminal
- Anthropic Agent sdk Metered Credits Openclaw Reinstated June 15 Splits Your Claude Subscription in two
Top AI Product 
Leave a comment