Shabana Khanam’s Post

Very cool new release by @OpenAI. Structured outputs for the new gpt-4o model. This means you can now force JSON response in strict mode AND provide a schema for it to adhere to. No more "mis-using" tool calls to force schema. https://lnkd.in/eRjv5hUP

OpenAI has introduced a new model named GPT-4O-mini. While many were expecting GPT-5 or GPT-4.5, it appears OpenAI has opted to change their strategy. The mini model is very affordable, with a knowledge base cutoff in October 2023. The API pricing for GPT-4O-mini is: - 15¢ per million tokens for input - 60¢ per million tokens for output - 128k context window This means GPT-4O-mini is 30 times cheaper and is comparable in performance to GPT-4. Additionally, GPT-3.5 has been discontinued and replaced by the mini model.

OpenAI has launched GPT-4o Mini: a high-performance, lightweight model that's both efficient and cost-effective. Priced at 15 cents per million input tokens and 60 cents per million output tokens (equivalent to 2500 book pages), GPT-4o Mini excels in reasoning tasks with an impressive MMLU of 82%. That's what claimed by OpenAI, Big techs are moving towards smaller models, so there's a tradeoff between efficiency and cost. For details see : https://lnkd.in/d7cJfCni

OpenAI's GPT-4o delivers output speed of 109 tokens/s in first independent benchmark This makes GPT-4o: • Over 5x faster than GPT-4 Turbo (~20 tokens/s) • Approximately twice as fast as GPT-3.5 (~55 tokens/s) This is a major leap forward in performance and cost for a frontier-class model. Artificial Analysis benchmarks models and providers 8x every day with a range of test workloads. Keep an eye on our Over Time charts to see whether OpenAI can maintain these speeds as developers shift workloads to the new model!

🔥 OpenAI is testing an alpha version of GPT-4o Long Output model.   🤔 This version implies 64k output token responses with GPT-4o, which would not only allow to get much longer answers, but also reducing costs by solving those scenarios where the prev answers needs to be re-introduced as part of the next prompt context in order to iterate and get longer codes/stories/answers.   🤓 Really interesting stuff to monitor Url: https://lnkd.in/d7xDG9af

OpenAI has launched an experimental version of GPT-4o capable of generating outputs up to 64,000 tokens per request. This new feature, accessible via the `gpt-4o-64k-output-alpha` model name, is designed to unlock new use cases requiring longer completions. Due to increased inference costs, the pricing for this model is higher: $6.00 per million input tokens and $18.00 per million output tokens. For more details, you can read the full announcement(https://lnkd.in/dCuwxvbG).

Great insights from Jonathan, as usual. The LLM ecosystem is hyper-competitive as each player tries to build more & more larger and responsive models. But, as these models get larger leading to higher energy requirements. It has created a new market for Small-Language Models (SLMs) that can be used for more specialized tasks and are better suited for app and on-device usages.

OpenAI just dropped their new flagship model - GPT-4o GPT-4o combines with function calling is insane. Here's an example. 1. User submits image of bugs/issues 2. GPT-4o to analyze it 3. Function calling to search for possible fixes Cheaper. Faster. What other use cases can you think of?

OpenAI just announced "GPT-4o mini", their new cost-efficient small model. Here are the main highlights (according to OpenAI, the link in the comments): 🟢 Smaller and more affordable than previous models, making it ideal for a wide range of applications. 🟢Performs well on a variety of reasoning tasks, including mathematical reasoning and coding. 🟢Built with safety measures to prevent misuse. 🟢GPT-4o mini is available now in the OpenAI API. ❌ But in the past 24 hours, the testers think otherwise. See their examples on the "OpenAI Community" website (link below). 🤦🏻♂️😐 The title of the page is quite self-explanatory: "𝐆𝐏𝐓 4𝐨 𝐦𝐢𝐧𝐢 𝐢𝐬 𝐝𝐮𝐦𝐛𝐞𝐫 𝐭𝐡𝐚𝐧 𝐲𝐨𝐮 𝐜𝐚𝐧 𝐭𝐡𝐢𝐧𝐤" I'm going to share some of the "real-world test results" in the comments. #AI #LLMs #SLMs #OpenAI #GPT4omini #DumbModels Denis O., Felix Hovsepian, Prof J. Mark Bishop, Walid Saba, Emmanuel Maggiori, Tim Scarfe, Axel C.

OpenAI released the price of GPT-4o mini which will enable a lot of business to use the LLM API more but it is still not cheap for big use cases. Here's a brief overview of the pricing for small LLM models from the main suppliers: 1. Claude Haiku Input : $0.25 / 1M token Output : $1.25 / 1M token 2. Llama-3 8b (Groq) Input : $0.05 / 1M token Output : $0.08 / 1M token 3. Gemini 1.5 Flash Input : $0.35 / 1M token Output : $1.05/ 1M token 4. GPT-4o mini Input : $0.15 / 1M token Output : $0.60 / 1M token For comparison, the standard GPT-4o pricing is : Input : $5.00 / 1M token Output : $15.00 / 1M token The price reduction is significant - but - keep in mind that using a Small Language Model impacts heavily output quality. My main concern with this price is knowing how computationally expensive llms are and the green factor for such big models are not that great. How are they making it available so cheap?