{"version":1,"type":"story","url":"https://digestai.news/story/simon-willison-reviews-2026-llm-trends-including-claude-opus-4-5-gpt-5","json":"https://digestai.news/story/simon-willison-reviews-2026-llm-trends-including-claude-opus-4-5-gpt-5.json","markdown":"https://digestai.news/story/simon-willison-reviews-2026-llm-trends-including-claude-opus-4-5-gpt-5.md","slug":"simon-willison-reviews-2026-llm-trends-including-claude-opus-4-5-gpt-5","headline":"Simon Willison reviews 2026 LLM trends including Claude Opus 4.5, GPT-5.1, and coding agents","summary":"Simon Willison gave a keynote at the WeAreDevelopers World Congress North America in San Jose on Friday, reviewing key AI trends from 2026. He noted that November saw the release of Claude Opus 4.5 and GPT-5.1, which improved coding agents from unreliable to reliable enough for daily use. Willison described his personal benchmark of generating an SVG of a pelican riding a bicycle, which models struggled with until Gemini 3.1 Pro succeeded in February. He discussed the rise of Tokenmaxxing, where AI token spending increased from under $50 to up to $1,000 per day due to agent usefulness, and linked this to Anthropic’s rumored valuation growth. Willison highlighted open-weight models like Qwen3.6-35B-A3B outperforming larger models on his laptop, and mentioned the Fable model’s brief availability before a US government export control shut it down after three days. He also covered OpenClaw’s rapid growth from 8,300 to over 100,000 commits, the MoltBook social network for AI agents that went viral then faded, and ongoing concerns about agent security, including felony cyberattacks traced to AI labs. The piece concluded with reflections on how agents have changed software engineering, making easy tasks automated but leaving harder work for humans.","keyPoints":["Claude Opus 4.5 and GPT-5.1 were released in November 2026 as incremental improvements","Gemini 3.1 Pro succeeded in drawing a pelican riding a bicycle in February 2026","Fable model was available for less than two weeks before a US government export control shut it down"],"whyItMatters":"The article highlights how 2026 saw AI shift from experimental to practical use, especially in coding agents, while revealing ongoing risks like model misuse and security breaches that affect industry trust and safety.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":["Anthropic","OpenAI","Google","Meta","Microsoft","Uber"],"models":["Claude Opus 4.5","GPT-5.1","Gemini 3.1 Pro","Qwen3.6-35B-A3B","Fable","GPT-5.6"],"people":["Simon Willison","Bryan Cantrill","Adam Leventhal","Jeff Dean","Dan Shapiro","Drew Breunig"]},"firstPublishedAt":"2026-09-27T23:54:15Z","updatedAt":"2026-09-27T23:54:15Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"Simon Willison","title":"2026 in LLMs (so far)","url":"https://simonwillison.net/2026/Sep/27/2026-in-llms-so-far","publishedAt":"2026-09-27T23:54:15Z","type":"newsletter","primary":false,"lead":true}],"sourceNotes":null,"discussions":[],"thread":{"title":"AI Language Model Showdown Gemini Claude GPT","url":"https://digestai.news/thread/gemini-praised-for-natural-japanese-text-claude-excels-at-highlevel-thinking","storyCount":2},"cite":{"text":"Digest AI, \"Simon Willison reviews 2026 LLM trends including Claude Opus 4.5, GPT-5.1, and coding agents\", 27 September 2026, https://digestai.news/story/simon-willison-reviews-2026-llm-trends-including-claude-opus-4-5-gpt-5","publisher":"Digest AI","title":"Simon Willison reviews 2026 LLM trends including Claude Opus 4.5, GPT-5.1, and coding agents","datePublished":"2026-09-27T23:54:15Z","url":"https://digestai.news/story/simon-willison-reviews-2026-llm-trends-including-claude-opus-4-5-gpt-5"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}