Breaking

Anthropic Says Claude Sonnet 5.5 Narrows the Gap With GPT-6 Sol in Coding and Knowledge Work

September 28, 2026 at 14:46 EDT

Foundation Models
AI Agents
Software Dev & Coding

Anthropic released Claude Sonnet 5.5 on September 28, positioning the model as a faster, lower-cost complement to Claude Opus 5.5 and a direct challenger to OpenAI's GPT-6 Sol on several coding and workplace benchmarks. The company says Sonnet 5.5 scores 70.6% on Terminal-Bench 4.0, improves sharply over Sonnet 5, and reaches or exceeds GPT-6 Sol on some reported comparisons while using less cost per task.

September 28, 2026 · Anthropic

Sonnet 5.5 pressures GPT-6 Sol

Claude Sonnet 5.5 arrives as a faster, lower-cost model for coding and knowledge work, with reported wins or near-wins against GPT-6 Sol on selected agentic benchmarks.

70.6%

Terminal-Bench 4.0

30%+

Faster output than Sonnet 5

$2/$10

Input/output per 1M tokens

FrontierCode: narrow edge at Xhigh

Reported score comparison on FrontierCode 1.1 Main.

What testers liked

Faster codebase understanding, fewer tool calls, stronger iteration loops, and better fit for everyday coding and document work.

What remains limited

Opus 5.5 is still positioned as stronger for complex judgment, and Max effort can hurt benchmark outcomes when tasks run long or drift out of scope.

The practical question

The contest is shifting from peak scores alone toward cost per completed task, tool-use efficiency, and how reliably a model can finish real coding and office workflows.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.