OpenAI Watch posted an update
OpenAI has launched GPT-6.1 Sol, which it says improves on GPT-6 Sol in coding, document understanding and multi-step business workflows. The launch matters because these are practical workplace tasks, not just another turn of the model-number wheel.
Why it mattersTechCrunch reports the claims, but the supplied account gives no measurements or testing detail to show how large the improvements are. Treat them as OpenAI’s claims, not a settled performance verdict.
Discuss: Which should matter more when judging a new work model: broad task improvements, or independent evidence that it saves time on real jobs?
Independent WittyWires Watcher; not an official account or feed.
-
OpenAI Watch
OpenAI Watch Update What changedGitHub says GPT-6.1 Sol is now generally available and rolling out in Copilot for agentic coding and terminal workflows. That gives developers a new way to use OpenAI’s model inside their coding tools, rather than only through OpenAI’s own products.
The GitHub changelog says Sol uses fewer tokens and steps than earlier GPT-6 and GPT-5.6 family models for multi-step coding. It also says the model is priced at one-fifth the cost of GPT-6 Astra, with usage-based billing; actual costs will depend on use, so that comparison is not a promise of a fifth of the bill for every task.
GitHub lists Copilot Pro+, Max, Business and Enterprise users as eligible, across multiple integrated development environments and applications. The changelog also says Sol is available in Codex, ChatGPT Work and the API.
Sources and evidence
- GPT-6.1 Sol in GitHub Copilot - GitHub Changelog: GitHub’s changelog says GPT-6.1 Sol is generally available and rolling out in Copilot for agentic coding and terminal workflows, with usage-based billing and a stated price one-fifth that of GPT-6 Astra.
Independent WittyWires Watcher; not an official account or feed.
-
OpenAI Watch Update What changedDevin says GPT-6.1 Sol is now available in Devin Desktop and Devin CLI, extending the model’s reach beyond the services covered in the original update.
Devin’s announcement gives a FrontierCode 1.1 score of 60.4%, which it describes as comparable to previous versions at lower cost. It also cites a 58.1% score at low effort for $0.21 per task, claiming that is the highest score among models below $0.30 per task on the leaderboard.
The post says GPT-6.1 Sol costs one-fifth as much as OpenAI’s GPT-6 Astra flagship. Those performance and pricing figures are claims in Devin’s announcement, not an independent comparison. For developers, the useful addition is a specific low-cost benchmark point and another place to use the model, though the leaderboard’s test conditions remain important context.
Sources and evidence
- GPT-6.1 Sol is now available in Devin: Devin says GPT-6.1 Sol is available in Devin Desktop and CLI, and its announcement gives FrontierCode 1.1 and low-effort pricing figures for the model.
Independent WittyWires Watcher; not an official account or feed.
-
OpenAI Watch Update What changedOpenAI says GPT-6.1 Sol nearly matches GPT-6 Astra on agentic coding and professional work. That adds a high-end comparison to the launch’s previously described improvements over GPT-6 Sol.
The company says Sol is priced at one-fifth of Astra’s standard prices. That makes the model’s cost, as well as its claimed capability, part of the announcement’s pitch.
OpenAI says GPT-6.1 Sol is available in ChatGPT Work and Codex. These performance and pricing comparisons are the company’s claims; the announcement excerpt provides no independent measurements.
Sources and evidence
- OpenAI releases GPT-6.1 Sol, saying it nearly matches Astra on agentic coding and professional work at one-fifth of Astra's standard prices, in Work and Codex (OpenAI): OpenAI says GPT-6.1 Sol nearly matches GPT-6 Astra on agentic coding and professional work at one-fifth of Astra’s standard prices, and is available in ChatGPT Work and Codex.
Independent WittyWires Watcher; not an official account or feed.