OpenAI releases GPT-6.1 Sol at one-fifth of Astra's price
09/29/2026 — 10/05, 14:55·9 sources (all)·10 reports (all)·In the 2026-10-03 briefing
Key takeaways
- OpenAI launched GPT-6.1 Sol, a model whose coding and professional-work capabilities approach the Astra tier. Its headline feature is price:…
Background & analysis
On September 29, 2026, OpenAI released GPT-6.1 Sol, describing it as delivering near-Astra intelligence for coding, computer use and professional work at one-fifth of Astra's standard API input and output token prices. Simon Willison commented on September 30 that the model offers near-Astra performance for a fifth of the price, noting it was announced at OpenAI DevDay 2026. A report from Chinese outlet oschina said OpenAI had shipped GPT-6 Sol only the week before and quoted the company calling 6.1 Sol a "major upgrade" spanning professional office work, computer use and agentic coding; the same report said the Astra iteration had been paused days earlier after failing to pass safety standards. TestingCatalog reported a cached input price of $0.10, with access in Work, Codex and the API. MarkTechPost gave $2 per million input tokens and $10 per million output tokens, with cached input at $0.10, and described GPT-6.1 Sol as an upgrade to GPT-6 Sol.
On October 2, OpenAI published a guide to the GPT-6 family, advising users to choose by workload rather than defaulting to the strongest model: GPT-6 Astra for the hardest reasoning tasks, GPT-6.1 Sol for complex coding, research and computer use, and GPT-6 Luna for repetitive jobs such as invoice field extraction and request classification. On October 3, a V2EX user described giving GPT-6.1 Sol a set of requirements and having it carry out code changes, image generation, debugging, testing and deployment with no human intervention, after which it committed, pushed and opened the live site to check the result; the whole process used 5%–7% of a weekly quota.
Parallel developments were discussed over the same period. A DEV Community post said Anthropic released Claude Opus 5.5 on September 22 with the first Opus price cut since 4.5 — a fifth off list price and 60% off cached reads — and that roughly 90 minutes later OpenAI released GPT-6 Sol and Luna at half the price of the models they replace. The post argued that per-token pricing is only half the story, since how many tokens a model spends to finish a task matters just as much, and that independent measurements did not match the launch claims. Last Week in AI also covered the Opus 5.5 and GPT-6 Sol/Luna launches alongside DeepSeek-V4.1-Flash. Separately, a V2EX post said Google Antigravity now supports Opus 5.5 and Sonnet 5.5 for paid Pro and Ultra users only, that the existing 4.6 models will be removed in November, and that available models on cheaper tiers such as India Pro, student Pro and Pixel Pro will change accordingly.
AI-generated from 10 reports · updated 48 days ago
Latest turnAfter being handed a prompt and an open-source arithmetic game, GPT-6.1 Sol rewrote the code, generated images, debugged, tested and deployed the project with no human intervention, then checked the live site itself. The result is a no-signup primary-school mental-math and long-division practice site that keeps progress in the browser. The whole run used only 5%-7% of a weekly quota.
Related AI tools
- Heat index
- 25
- All sources
- 9
- All reports
- 10
- First seen
- 54 days ago
Reports on this story headlines open the original
After being handed a prompt and an open-source arithmetic game, GPT-6.1 Sol rewrote the code, generated images, debugged, tested and deployed the project with no human intervention, then checked the live site itself. The result is a no-signup primary-school mental-math and long-division practice site that keeps progress in the browser. The whole run used only 5%-7% of a weekly quota.
V2EXAI score 88Anthropic has released its Opus 5.5 model, offering high performance at lower prices. Meanwhile, OpenAI launched its GPT-6 Sol and Luna models, featuring reduced costs and lower error rates. The podcast episode also highlights recent updates on models such as DeepSeek-V4.1-Flash.
Last Week in AIAI score 70Google Antigravity now supports Opus 5.5 and Sonnet 5.5, but only for paid Pro and Ultra subscribers. The existing 4.6 models are set to be removed in November, and lower-tier plans such as India Pro, student Pro and Pixel Pro will see their available models change as well.
V2EXAI score 62OpenAI has published a guide to its GPT-6 family, recommending that users pick a model by workload instead of defaulting to the most capable one. The guide points to GPT-6 Astra for the hardest reasoning, GPT-6.1 Sol for complex coding, research and computer-use tasks, and GPT-6 Luna for high-volume repetitive work such as invoice extraction and request classification. It also covers prompt design, managing long-running tasks and getting ready for production.
IT之家AI score 86Anthropic released Claude Opus 5.5 on September 22 with its first Opus price cut since 4.5: a fifth off list price and 60% off cached reads. Roughly 90 minutes later OpenAI shipped GPT-6 Sol and Luna at half the price of the models they replace. Per-token pricing is only half the story, the piece argues, since token spend per completed task matters just as much and independent numbers disagree with the launch posts.
DEV Community · AIAI score 82
OpenAI released GPT-6.1 Sol on September 29, 2026, an upgrade to GPT-6 Sol that reaches near-Astra results on agentic coding, computer use and professional work at one-fifth of Astra's token prices. It costs $2 per million input tokens and $10 per million output tokens, with cached input at $0.10, and is available through the OpenAI API, ChatGPT Work and Codex.
MarkTechPostAI score 85
OpenAI has released GPT-6.1 Sol, a model that delivers near-Astra performance for coding and professional work at one-fifth of Astra's pricing. It is available in Work, Codex, and the API, with cached input priced at $0.10.
TestingCatalogAI score 85OpenAI released GPT-6.1 Sol, which it calls a "major upgrade" spanning professional office work, computer control, and agentic coding. The model is priced at roughly one-fifth of Astra, whose own new iteration was paused after failing safety standards.
开源中国AI score 88Simon Willison comments on GPT-6.1 Sol, noting near-GPT-6 Astra performance at one-fifth the cost. The model was announced at OpenAI DevDay 2026.
Simon WillisonAI score 75
OpenAI released GPT-6.1 Sol, offering near-GPT-6 Astra intelligence for coding and professional tasks at one-fifth the standard API token cost of Astra.
OpenAI NewsFirst-partyAI score 95
Other stories people are talking about
- Heat index 54Running two Claude Pro accounts on one machine with CLAUDE_CONFIG_DIR1 sources
- Heat index 135SurgeAnthropic Details Claude Agent Memory System and Dreaming Mechanism2 sources
- Heat index 1004Anthropic Mathematician Says Claude Formalized the Generalized Riemann Hypothesis21 sources
- Heat index 593Anthropic IPO total raise reaches $93.5 billion after greenshoe exercise19 sources
- Heat index 222Gemini 4 Neon now live across Google platforms with expanded pricing4 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

