It's another Daft Punk day. Change a string in your program* and it's better, faster, cheaper: pick 3.
*Then fix all your prompts over the next two weeks.
GPT-4.1 in the API
251–260 of 513 posts
Re: GPT-4.1 in the API
#252Sam Altman wrote in February that GPT-4.5 would be "our last non-chain-of-thought model" [1], but GPT-4.1 also does not have internal chain-of-thought [2]. It seems like OpenAI keeps changing its plans. Deprecating GPT-4.5 less than 2 months after introducing it also seems unlikely to be the original plan. Changing plans is necessarily a bad thing, but I wonder why. Did they not expect this model to turn out as well…
Maybe that's why they named this model 4.1, despite coming out after 4.5 and supposedly outperforming it. They can pretend GPT-4.5 is the last non-chain-of-thought model by just giving all non-chain-of-thought-models version numbers below 4.5
Re: GPT-4.1 in the API
#253don't miss that OAI also published a prompting guide WITH RECEIPTS for GPT 4.1 specifically for those building agents... with a new recommendation for: - telling the model to be persistent (+20%) - dont self-inject/parse toolcalls (+2%) - prompted planning (+4%) - JSON BAD - use XML or arxiv 2406.13121 (GDM format) - put instructions + user query at TOP -and- BOTTOM - bottom-only is VERY BAD - no evidence that ALL CA…
>- dont self-inject/parse toolcalls (+2%)
What is meant by this?
Re: GPT-4.1 in the API
#2544.10 > 4.5 — @stevenheidel @sama: underrated tweet Source: https://x.com/stevenheidel/status/1911833398588719274
Too bad OpenAI named it 4.1 instead of 4.10. You can either claim 4.10 > 4.5 (the dots separate natural numbers) or 4.1 == 4.10 (they are decimal numbers), but you can't have both at once
Re: GPT-4.1 in the API
#255> They feature a refreshed knowledge cutoff of June 2024. As opposed to Gemini 2.5 Pro having cutoff of Jan 2025. Honestly this feels underwhelming and surprising. Especially if you're coding with frameworks with breaking changes, this can hurt you.
It's definitely an issue. Even the simplest use case of "create React app with Vite and Tailwind" is broken with these models right now because they're not up to date.
Try getting then to output Svelte 5 code...
Re: GPT-4.1 in the API
#256Is the version number a retcon of 4.5? On OpenAI's models page the names appear completely reasonable [1]: The o1 and o3 reasoning models, and non-reasoning there is 3.5, 4, 4o and 4.1 (let's pretend 4o makes sense). But that is only reasonable as long as we pretend 4.5 never happened, which the models page apparently does
Re: GPT-4.1 in the API
#257don't miss that OAI also published a prompting guide WITH RECEIPTS for GPT 4.1 specifically for those building agents... with a new recommendation for: - telling the model to be persistent (+20%) - dont self-inject/parse toolcalls (+2%) - prompted planning (+4%) - JSON BAD - use XML or arxiv 2406.13121 (GDM format) - put instructions + user query at TOP -and- BOTTOM - bottom-only is VERY BAD - no evidence that ALL CA…
> - JSON BAD - use XML or arxiv 2406.13121 (GDM format) And yet, all function calling and MCP is done through JSON...
Yeah anyone who has worked with these models knows how much they struggle with JSON inputs.
Re: GPT-4.1 in the API
#258Sam Altman wrote in February that GPT-4.5 would be "our last non-chain-of-thought model" [1], but GPT-4.1 also does not have internal chain-of-thought [2]. It seems like OpenAI keeps changing its plans. Deprecating GPT-4.5 less than 2 months after introducing it also seems unlikely to be the original plan. Changing plans is necessarily a bad thing, but I wonder why. Did they not expect this model to turn out as well…
Why do you expect to believe a single word Sam Altman says?
Re: GPT-4.1 in the API
#259More information here:
https://platform.openai.com/docs/models/gpt-4.1
https://platform.openai.com/docs/models/gpt-4.1-mini
https://platform.openai.com/docs/models/gpt-4.1-nanoRe: GPT-4.1 in the API
#260Earlier quoted context omitted.
The problem is that while you can train a model with the hyperparameter of "context size" set to 1M, there's very little 1M data to train on. Most of your model's ability to follow long context comes from the fact that it's trained on lots of (stolen) books; in fact I believe OpenAI just outright said in court that they can't do long context without training on books. Novels are usually measured in terms of words; an…
I'm not sure to which extent this opinion is accurately informed. It is well known that nobody trains on 1M token-long content. It wouldn't work anyway as the dependencies are too far fetched and you end up with vanishing gradients. RoPE (Rotary Positional Embeddings, think modulo or periodic arithmetics) scaling is key, whereby the model is trained on 16k tokens long content, and then scaled up to 100k+ [0]. Qwen 1M…
But Llama 4 Scout does badly on long context benchmarks despite claiming 10M. It scores 1 slot above Llama 3.1 8B in this one[1].