GPT-6 Sol vs Claude Sonnet 5

Since September 22, 2026 these two carry the same list price, $2 per million input tokens and $10 output, with cached input at $0.20 on both, so the rate card settles nothing. GPT-6 Sol is OpenAI's new mid tier, launched that day at half the price of GPT-5.6 Sol. Claude Sonnet 5 has been in production since June 30, 2026, and $2/$10 became its standard rate when Anthropic cancelled a planned step to $3/$15. On the only benchmark in our data that an independent party ran on both, Sol is clearly ahead: Cognition's FrontierCode v1.1 Main scores it 49.3 percent at max effort against 42.7 for Sonnet 5. Sol also has independent aggregate placements, eighth on the Vals Index at 62.57 percent and 48 on the Artificial Analysis Intelligence Index; we have no verified Vals Index figure for Sonnet 5 to set beside them. Sonnet 5's record is older and deeper. Vals AI measured it at 87.55 on MMLU-Pro and 88.89 on GPQA Diamond before archiving both benchmarks on September 1, and Anthropic reports 85.2 on SWE-bench Verified and 84.7 on BrowseComp, columns OpenAI left empty for Sol. The windows are close, 1,050,000 tokens for Sol against 1,000,000 for Sonnet 5, but OpenAI bills any Sol request above 272,000 input tokens at 2x input and cache and 1.5x output for the whole request, which is $4 and $15, so a long-context budget needs its own line. Beyond that the split is ecosystem and track record: Sonnet 5 runs on the Claude API, Amazon Bedrock, Google Vertex, and Microsoft Foundry and slots into Claude Code and MCP stacks, while Sol ships through the OpenAI API, Codex, ChatGPT Work, and Azure Foundry and is days old.