A breakthrough could literally be a single word. UW + Stanford found telling their model to “wait” during reasoning improved answer accuracy https://
arxiv.org/pdf/2501.19393
PROMPT ENGINEERING
-
Wait Instruction Improves AI Model Reasoning Accuracy
By
–
-
Google AI Studio adds branching to Gemini models
By
–
Google AI Studio now supports conversation branching on Gemini models. https://t.co/5XoSIHyW3j pic.twitter.com/OyAOHzxFEI
— 🚨 AI News | TestingCatalog (@testingcatalog) 24 février 2025Google AI Studio now supports conversation branching on Gemini models.
-
Using Sonnet 3.7 API Avoid Claude Rate Limits
By
–
If you talk to Claude a lot & get rate limited you can use Sonnet 3.7 with no usage limits via the API.
— Aaron Ng (@localghost) 24 février 2025
You can even attach the Claude System Prompt (available on their website) to get the personality right. pic.twitter.com/28RZmhIcQkIf you talk to Claude a lot & get rate limited you can use Sonnet 3.7 with no usage limits via the API. You can even attach the Claude System Prompt (available on their website) to get the personality right.
-
Claude 3.7 Sonnet et Cursor créent une page Airbnb complète
By
–
omg… 3.7 sonnet + latest cursor update (agent mode)
— Ben Tossell (@bentossell) 24 février 2025
one-prompt (+2 'errors') to this airbnb clone – full listing page and filtering
literally just accept all and watch it do its thing https://t.co/07hX7krKEX pic.twitter.com/VorJJDJluSomg… 3.7 sonnet + latest cursor update (agent mode) one-prompt (+2 'errors') to this airbnb clone – full listing page and filtering literally just accept all and watch it do its thing
-
Cursor Update: Unified Chat and Composer with Improved Intelligence
By
–
latest cursor update has: – chat + composer in one (you can switch mid-convo)
– feels smarter too -
Claude 3.7 Sonnet Counts ‘r’s in ‘strawberry’
By
–
Claude 3.7 Sonnet with Thinking demo 👀
— 🚨 AI News | TestingCatalog (@testingcatalog) 24 février 2025
It made an artefact to count how many r's in the word "strawberry", so I can click to count them 😎 https://t.co/RWKPIRf2Xz pic.twitter.com/ybiqIovUxoClaude 3.7 Sonnet with Thinking demo It made an artefact to count how many r's in the word "strawberry", so I can click to count them
-
Control Claude’s Thinking Budget with New API Beta Header
By
–
When using our API, you have control over Claude's thinking budget, letting you balance speed/cost with answer quality. With our new beta header, you can let Claude think/output up to 128k tokens.
-
Sonnet-3.7 Available in Cursor for Advanced Coding
By
–
Sonnet-3.7 is available in Cursor!
— Cursor (@cursor_ai) 24 février 2025
We've been very impressed by its coding ability, especially on real-world agentic tasks. It appears to be the new state of the art. pic.twitter.com/PisIs2X9MbSonnet-3.7 is available in Cursor! We've been very impressed by its coding ability, especially on real-world agentic tasks. It appears to be the new state of the art.
-
Claude Sonnet 3.7 excels in SWE benchmarks
By
–
Claude Sonnet 3.7 evals are here too A huge jump on the SWE bench
-
Questioning if AI responses are simply template-based conditionals
By
–
How do we know it's not simply something like this “`
name = …
say_thanks = True
print(f"Hi {name}. To ask Gemini to reference past chats, please ensure… {'Thanks.' if say_thanks else ' '}")
“`