Anthropic发布Claude Sonnet
Claude Sonnet 5.5
Sonnet系列是主流轻量级旗舰,这次不仅性能跃迁还直接接管了免费层,对开发者选型和竞品格局都有直接影响,值得关注。
Claude Sonnet 5.5
New Sonnet model from Anthropic today. They say it "runs 30%+ faster, and costs up to 30% less for most work" - it's priced the same as Sonnet 5 but appears to beat it on every benchmark, and should be cheaper to run as well.
Anthropic 今天发布了新的 Sonnet 模型。他们称其“运行速度快 30% 以上,大多数工作成本降低高达 30%”——其定价与 Sonnet 5 相同,但在各项基准测试中似乎都优于 Sonnet 5,并且运行成本也应更低。
Here are some pelicans riding bicycles. Sonnet 5.5 suffered from the same bug as Opus 5.5: the "max" thinking effort pelican thought for 128,000 tokens (at a cost of $1.28) before running out of tokens and failing to produce an SVG.
这里有一些骑自行车的鹈鹕。Sonnet 5.5 出现了与 Opus 5.5 相同的 bug:在“max”思考模式下,这只鹈鹕在耗尽令牌(token)并未能生成 SVG 之前,思考了 128,000 个令牌(花费 1.28 美元)。
Here's the pelican it gave me for thinking effort "xhigh", at a cost of 5.74 cents and taking 41 seconds:
这是我在“xhigh”思考模式下得到的鹈鹕,花费 5.74 美分,耗时 41 秒:
Sonnet 5.5 appears to be almost as good as Opus 5.5 on some coding tasks, including various viral 3D animation tricks.
在一些编码任务中,包括各种病毒式传播的 3D 动画技巧,Sonnet 5.5 的表现几乎与 Opus 5.5 相当。
The most interesting thing about Sonnet 5.5 is that it's now the model used for the free tier on claude.ai. OpenAI's ChatGPT free tier uses Luna 5.6, which means Anthropic currently have a much more capable free offering.
关于 Sonnet 5.5 最有趣的一点是,它现在被用于 claude.ai 的免费层级。OpenAI 的 ChatGPT 免费层级使用 Luna 5.6,这意味着 Anthropic 目前拥有能力更强的免费服务。
I ran this prompt against that free tier:
我对该免费层级运行了以下提示词:
build me an HTML page that renders a three-dimensional pelican riding a bicycle using WebGL
用 WebGL 为我构建一个 HTML 页面,渲染一只骑自行车的三维鹈鹕
And got back this page, which is a solid effort.
并得到了这个页面,这是一个扎实的成果。
Anthropic's announcement reiterates that Haiku 5.5 will be available "in the coming weeks". I really hope that one is price-competitive with GPT-6 Luna!
Anthropic 的公告重申 Haiku 5.5 将在“未来几周内”可用。我真的希望那个模型在价格上能与 GPT-6 Luna 竞争!
Tags: ai, generative-ai, llms, anthropic, claude, pelican-riding-a-bicycle, llm-release
标签:ai, generative-ai, llms, anthropic, claude, pelican-riding-a-bicycle, llm-release
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力