Claude Sonnet 5.5 vs GPT-6.1 Sol: Which AI Model Is Worth Using?

Two AI giants launch new models, one day apart, at the same price

On September 28 and 29, 2026, Anthropic and OpenAI each rolled out a new flagship model: Claude Sonnet 5.5 and GPT-6.1 Sol. In an interesting coincidence, both landed with identical pricing: $2 per million input tokens and $10 per million output tokens. Same budget, very different strengths. This post breaks down what each model does well, so you can pick the right one for your work, whether that's writing code, automating tasks, or everyday Q&A.

Claude Sonnet 5.5: a big step up for coding and computer use

Sonnet 5.5 is an upgrade to Sonnet 5, arriving with a clear promise: over 30% faster, up to 30% cheaper per task, while keeping the same list price as before.

The standout feature is its ability to work directly in the terminal, issuing commands and carrying out technical operations rather than just answering questions. On Terminal Bench, Sonnet 5.5 scored 70.6%, compared to just 10.3% for the older Sonnet 5, and it even beat Claude Opus 5.5, Anthropic's pricier flagship model (which scored 66.4%). In other words, you don't need to pay extra for Opus to get strong coding performance, at least not on this benchmark.

Sonnet 5.5 is also leading three other rankings: writing apps from plain-language descriptions (Vibe Code Bench), converting code between programming languages (Code Migration), and solving complex biology puzzles (BioMysteryBench). See the full breakdown in this article on Claude Sonnet 5.5.

GPT-6.1 Sol: near flagship power at a fifth of the price

On OpenAI's side, GPT-6.1 Sol was unveiled right at DevDay 2026 with a clear pitch: performance close to GPT-6 Astra, OpenAI's top-tier model, at roughly one-fifth the price (Astra costs $10 and $50 per million input and output tokens).

On the DeepSWE software-debugging benchmark, Sol matched Astra's score at a fifth of the cost. On OSWorld, a test of automated computer use, Sol trailed Astra by just 2.1 points while costing about seven times less. Notably, on AutomationBench, which measures how well an AI can carry out multi-step workflows on its own, Sol actually edged out Anthropic's Opus 5.5 by 2.2 percentage points. On top of that, reusing previously cached content cuts the price by up to 95%, down to just $0.10 per million tokens. Read more in this roundup of DevDay 2026 and GPT-6.1 Sol.

So which one is actually stronger?

No benchmark has pitted these two models directly against each other yet, so the honest answer is: it depends on what you need to do.

• If your work revolves around writing code, debugging, or working from the command line, Sonnet 5.5 delivers genuinely impressive results, even outperforming Anthropic's own pricier Opus 5.5.
• If you need AI to handle multi-step automation, control a computer screen, or process large workloads on a tight budget, GPT-6.1 Sol is well worth considering, since it comes close to OpenAI's top model at a fraction of the cost.

Both releases point to a broader 2026 trend: AI companies are no longer racing to build only the most powerful model, they're racing to build one that's "powerful and affordable" enough for everyone, not just large businesses.

Which model should you pick?

If you're new to AI and still not sure what Claude actually is or how to use it, start with this beginner's guide to Claude AI before diving into deeper comparisons. If you already use AI to help with coding, the fastest way to find your fit is to try both models on your real work for a week, since benchmark scores are only a reference point, what matters most is how each one actually feels in practice.