What shipped |
● OpenAI · GPT · Sep 22 Half the price of GPT-5.6 Sol with about half as many factual mistakes, while its all-round score rises only one point. Compared with GPT-5.6 Sol · the model it replaces about half ▲ vs GPT-5.6 Sol · OpenAI internal test Makes about half as many factual mistakes · OpenAI factuality test | $2 / $10 ▼ down from $4 / $20 · 50% cheaper Costs half as much as the model it replaces · Price per 1M tokens |
|
● Anthropic · Claude Opus · Sep 22 Anthropic's new Opus scores 7 points higher than Opus 5 on the all-round index and costs 20% less per token. Compared with Claude Opus 5 · the model it replaces 66.4% ▲ up from 52.3% · +14.1 points Completes two in three hard command-line tasks · Terminal-Bench 4.0 | $4 / $20 ▼ down from $5 / $25 · 20% cheaper Costs a fifth less than the model it replaces · Price per 1M tokens |
|
● OpenAI · GPT Luna · Sep 22 Half the input price and 58% less on output than GPT-5.6 Luna, with the same all-round score. Compared with GPT-5.6 Luna · the model it replaces 66.6% ✦ no GPT-5.6 Luna score published to compare Solves two in three hard coding tasks · DeepSWE 1.1 | $0.10 / $0.50 ▼ down from $0.20 / $1.20 · output 58% cheaper Costs about half as much as the model it replaces · Price per 1M tokens |
|
Also new this week |
● xAI · Grok · Sep 21 A larger base model than Grok 4.6, trained longer on multi-hour agentic tasks, at the same $2 / $6 price and the same 500k-token context. |
● Alibaba · Qwen-Image · image model · open weights · released Sep 20 · added this week A 7B open-weight model that unifies generation and editing, adds native transparent (RGBA) output and up to 10 reference images, under a research-only license. |
● Alibaba · Qwen Max & Plus · released Sep 2 · added this week An upgraded snapshot of Qwen 3.8-Max, further trained for coding and office-style agent work, with the same 2.4 trillion parameters, 1M-token context window and $2 / $6 price. |
Coming next |
Close to their usual gap Time since the last release is close to the usual gap between this line’s models. We don’t predict dates. |
90 days since the last release — usual gap is 98 |
● Muse · Meta · last: Muse Spark 1.3 26 days since the last release — usual gap is 28 |
|
Next AI Model checks 9 AI makers for new models every morning. Every number comes from the maker or an independent test, with its source. One email a week, only when something ships. |
