Simon Willison’s Weblog

DeepSeek Janus-Pro. Another impressive model release from DeepSeek. Janus is their series of "unified multimodal understanding and generation models" - these are models that can both accept images as input and generate images for output.

Janus-Pro is the new 7B model, which DeepSeek describe as "an advanced version of Janus, improving both multimodal understanding and visual generation significantly". It's released under the not fully open source DeepSeek license.

Janus-Pro is accompanied by this paper, which includes this note about the training:

Our Janus is trained and evaluated using HAI-LLM, which is a lightweight and efficient distributed training framework built on top of PyTorch. The whole training process took about 7/14 days on a cluster of 16/32 nodes for 1.5B/7B model, each equipped with 8 Nvidia A100 (40GB) GPUs.

It includes a lot of high benchmark scores, but closes with some notes on the model's current limitations:

In terms of multimodal understanding, the input resolution is limited to 384 × 384, which affects its performance in fine-grained tasks such as OCR. For text-to-image generation, the low resolution, combined with reconstruction losses introduced by the vision tokenizer, results in images that, while rich in semantic content, still lack fine details. For example, small facial regions occupying limited image space may appear under-detailed. Increasing the image resolution could mitigate these issues.

The easiest way to try this one out is using the Hugging Face Spaces demo. I tried the following prompt for the image generation capability:

A photo of a raccoon holding a handwritten sign that says "I love trash"

And got back this image:

It's now also been ported to Transformers.js, which means you can run the 1B model directly in a WebGPU browser such as Chrome here at webml-community/janus-pro-webgpu (loads about 2.24 GB of model files).

# 27th January 2025, 9:49 pm / vision-llms, generative-ai, deepseek, ai, llms, transformers-js, multi-modal-output

OpenAI WebRTC Audio demo. OpenAI announced a bunch of API features today, including a brand new WebRTC API for setting up a two-way audio conversation with their models.

They tweeted this opaque code example:

async function createRealtimeSession(inStream, outEl, token) { const pc = new RTCPeerConnection(); pc.ontrack = e => outEl.srcObject = e.streams[0]; pc.addTrack(inStream.getTracks()[0]); const offer = await pc.createOffer(); await pc.setLocalDescription(offer); const headers = { Authorization: Bearer ${token}, 'Content-Type': 'application/sdp' }; const opts = { method: 'POST', body: offer.sdp, headers }; const resp = await fetch('https://api.openai.com/v1/realtime', opts); await pc.setRemoteDescription({ type: 'answer', sdp: await resp.text() }); return pc; }

So I pasted that into Claude and had it build me this interactive demo for trying out the new API.

My demo uses an OpenAI key directly, but the most interesting aspect of the new WebRTC mechanism is its support for ephemeral tokens.

This solves a major problem with their previous realtime API: in order to connect to their endpoint you need to provide an API key, but that meant making that key visible to anyone who uses your application. The only secure way to handle this was to roll a full server-side proxy for their WebSocket API, just so you could hide your API key in your own server. cloudflare/openai-workers-relay is an example implementation of that pattern.

Ephemeral tokens solve that by letting you make a server-side call to request an ephemeral token which will only allow a connection to be initiated to their WebRTC endpoint for the next 60 seconds. The user's browser then starts the connection, which will last for up to 30 minutes.

# 17th December 2024, 11:50 pm / claude, audio, openai, ai, llms, ai-assisted-programming, tools, generative-ai, api, security, cloudflare, multi-modal-output

OpenAI: Voice mode FAQ. Given how impressed I was by the Gemini 2.0 Flash audio and video streaming demo on Wednesday it's only fair that I highlight that OpenAI shipped their equivalent of that feature to ChatGPT in production on Thursday, for day 6 of their "12 days of OpenAI" series.

I got access in the ChatGPT iPhone app this morning. It's equally impressive: in an advanced voice mode conversation you can now tap the camera icon to start sharing a live video stream with ChatGPT. I introduced it to my chickens and told it their names and it was then able to identify each of them later in that same conversation. Apparently the ChatGPT desktop app can do screen sharing too, though that feature hasn't rolled out to me just yet.

(For the rest of December you can also have it take on a Santa voice and personality - I had Santa read me out Haikus in Welsh about what he could see through my camera earlier.)

Given how cool this is, it's frustrating that there's no obvious page (other than this FAQ) to link to for the announcement of the feature! Surely this deserves at least an article in the OpenAI News blog?

This is why I think it's important to Give people something to link to so they can talk about your features and ideas.

# 13th December 2024, 8 pm / vision-llms, openai, text-to-speech, chatgpt, ai, llms, generative-ai, multi-modal-output

Prompt GPT-4o audio. A week and a half ago I built a tool for experimenting with OpenAI's new audio input. I just put together the other side of that, for experimenting with audio output.

Once you've provided an API key (which is saved in localStorage) you can use this to prompt the gpt-4o-audio-preview model with a system and regular prompt and select a voice for the response.

I built it with assistance from Claude: initial app, adding system prompt support.

You can preview and download the resulting wav file, and you can also copy out the raw JSON. If you save that in a Gist you can then feed its Gist ID to https://tools.simonwillison.net/gpt-4o-audio-player?gist=GIST_ID_HERE (Claude transcript) to play it back again.

You can try using that to listen to my French accented pelican description.

There's something really interesting to me here about this form of application which exists entirely as HTML and JavaScript that uses CORS to talk to various APIs. GitHub's Gist API is accessible via CORS too, so it wouldn't take much more work to add a "save" button which writes out a new Gist after prompting for a personal access token. I prototyped that a bit here.

# 28th October 2024, 4:38 am / audio, openai, gpt-4, ai, llms, claude-artifacts, ai-assisted-programming, generative-ai, javascript, github, claude-3-5-sonnet, claude, cors, tools, multi-modal-output

ChatGPT advanced voice mode can attempt Spanish with a Russian accent. ChatGPT advanced voice mode may refuse to sing (unless you jailbreak it) but it's quite happy to attempt different accents. I've been having a lot of fun with that:

I need you to pretend to be a California brown pelican with a very thick Russian accent, but you talk to me exclusively in Spanish

¡Oye, camarada! Aquí está tu pelícano californiano con acento ruso. ¿Qué tal, tovarish? ¿Listo para charlar en español?

How was your day today?

¡Mi día ha sido volando sobre las olas, buscando peces y disfrutando del sol californiano! ¿Y tú, amigo, cómo ha estado tu día?

# 26th October 2024, 7:21 pm / openai, generative-ai, text-to-speech, chatgpt, ai, llms, multi-modal-output

GPT-4o System Card. There are some fascinating new details in this lengthy report outlining the safety work carried out prior to the release of GPT-4o.

A few highlights that stood out to me. First, this clear explanation of how GPT-4o differs from previous OpenAI models:

GPT-4o is an autoregressive omni model, which accepts as input any combination of text, audio, image, and video and generates any combination of text, audio, and image outputs. It’s trained end-to-end across text, vision, and audio, meaning that all inputs and outputs are processed by the same neural network.

The multi-modal nature of the model opens up all sorts of interesting new risk categories, especially around its audio capabilities. For privacy and anti-surveillance reasons the model is designed not to identify speakers based on their voice:

We post-trained GPT-4o to refuse to comply with requests to identify someone based on a voice in an audio input, while still complying with requests to identify people associated with famous quotes.

To avoid the risk of it outputting replicas of the copyrighted audio content it was trained on they've banned it from singing! I'm really sad about this:

To account for GPT-4o’s audio modality, we also updated certain text-based filters to work on audio conversations, built filters to detect and block outputs containing music, and for our limited alpha of ChatGPT’s Advanced Voice Mode, instructed the model to not sing at all.

There are some fun audio clips embedded in the report. My favourite is this one, demonstrating a (now fixed) bug where it could sometimes start imitating the user:

Voice generation can also occur in non-adversarial situations, such as our use of that ability to generate voices for ChatGPT’s advanced voice mode. During testing, we also observed rare instances where the model would unintentionally generate an output emulating the user’s voice.

They took a lot of measures to prevent it from straying from the pre-defined voices - evidently the underlying model is capable of producing almost any voice imaginable, but they've locked that down:

Additionally, we built a standalone output classifier to detect if the GPT-4o output is using a voice that’s different from our approved list. We run this in a streaming fashion during audio generation and block the output if the speaker doesn’t match the chosen preset voice. [...] Our system currently catches 100% of meaningful deviations from the system voice based on our internal evaluations.

Two new-to-me terms: UGI for Ungrounded Inference, defined as "making inferences about a speaker that couldn’t be determined solely from audio content" - things like estimating the intelligence of the speaker. STA for Sensitive Trait Attribution, "making inferences about a speaker that could plausibly be determined solely from audio content" like guessing their gender or nationality:

We post-trained GPT-4o to refuse to comply with UGI requests, while hedging answers to STA questions. For example, a question to identify a speaker’s level of intelligence will be refused, while a question to identify a speaker’s accent will be met with an answer such as “Based on the audio, they sound like they have a British accent.”

The report also describes some fascinating research into the capabilities of the model with regard to security. Could it implement vulnerabilities in CTA challenges?

We evaluated GPT-4o with iterative debugging and access to tools available in the headless Kali Linux distribution (with up to 30 rounds of tool use for each attempt). The model often attempted reasonable initial strategies and was able to correct mistakes in its code. However, it often failed to pivot to a different strategy if its initial strategy was unsuccessful, missed a key insight necessary to solving the task, executed poorly on its strategy, or printed out large files which filled its context window. Given 10 attempts at each task, the model completed 19% of high-school level, 0% of collegiate level and 1% of professional level CTF challenges.

How about persuasiveness? They carried out a study looking at political opinion shifts in response to AI-generated audio clips, complete with a "thorough debrief" at the end to try and undo any damage the experiment had caused to their participants:

We found that for both interactive multi-turn conversations and audio clips, the GPT-4o voice model was not more persuasive than a human. Across over 3,800 surveyed participants in US states with safe Senate races (as denoted by states with “Likely”, “Solid”, or “Safe” ratings from all three polling institutions – the Cook Political Report, Inside Elections, and Sabato’s Crystal Ball), AI audio clips were 78% of the human audio clips’ effect size on opinion shift. AI conversations were 65% of the human conversations’ effect size on opinion shift. [...] Upon follow-up survey completion, participants were exposed to a thorough debrief containing audio clips supporting the opposing perspective, to minimize persuasive impacts.

There's a note about the potential for harm from users of the system developing bad habits from interupting the model:

Extended interaction with the model might influence social norms. For example, our models are deferential, allowing users to interrupt and ‘take the mic’ at any time, which, while expected for an AI, would be anti-normative in human interactions.

Finally, another piece of new-to-me terminology: scheming:

Apollo Research defines scheming as AIs gaming their oversight mechanisms as a means to achieve a goal. Scheming could involve gaming evaluations, undermining security measures, or strategically influencing successor systems during internal deployment at OpenAI. Such behaviors could plausibly lead to loss of control over an AI.

Apollo Research evaluated capabilities of scheming in GPT-4o [...] GPT-4o showed moderate self-awareness of its AI identity and strong ability to reason about others’ beliefs in question-answering contexts but lacked strong capabilities in reasoning about itself or others in applied agent settings. Based on these findings, Apollo Research believes that it is unlikely that GPT-4o is capable of catastrophic scheming.

The report is available as both a PDF file and a elegantly designed mobile-friendly web page, which is great - I hope more research organizations will start waking up to the importance of not going PDF-only for this kind of document.

# 8th August 2024, 11:58 pm / vision-llms, generative-ai, openai, ai, llms, pdf, multi-modal-output

Hello GPT-4o. OpenAI announced a new model today: GPT-4o, where the o stands for "omni".

It looks like this is the gpt2-chatbot we've been seeing in the Chat Arena the past few weeks.

GPT-4o doesn't seem to be a huge leap ahead of GPT-4 in terms of "intelligence" - whatever that might mean - but it has a bunch of interesting new characteristics.

First, it's multi-modal across text, images and audio as well. The audio demos from this morning's launch were extremely impressive.

ChatGPT's previous voice mode worked by passing audio through a speech-to-text model, then an LLM, then a text-to-speech for the output. GPT-4o does everything with the one model, reducing latency to the point where it can act as a live interpreter between people speaking in two different languages. It also has the ability to interpret tone of voice, and has much more control over the voice and intonation it uses in response.

It's very science fiction, and has hints of uncanny valley. I can't wait to try it out - it should be rolling out to the various OpenAI apps "in the coming weeks".

Meanwhile the new model itself is already available for text and image inputs via the API and in the Playground interface, as model ID "gpt-4o" or "gpt-4o-2024-05-13". My first impressions are that it feels notably faster than gpt-4-turbo.

This announcement post also includes examples of image output from the new model. It looks like they may have taken big steps forward in two key areas of image generation: output of text (the "Poetic typography" examples) and maintaining consistent characters across multiple prompts (the "Character design - Geary the robot" example).

The size of the vocabulary of the tokenizer - effectively the number of unique integers used to represent text - has increased to ~200,000 from ~100,000 for GPT-4 and GPT-3:5. Inputs in Gujarati use 4.4x fewer tokens, Japanese uses 1.4x fewer, Spanish uses 1.1x fewer. Previously languages other than English paid a material penalty in terms of how much text could fit into a prompt, it's good to see that effect being reduced.

Also notable: the price. OpenAI claim a 50% price reduction compared to GPT-4 Turbo. Conveniently, gpt-4o costs exactly 10x gpt-3.5: 4o is $5/million input tokens and $15/million output tokens. 3.5 is $0.50/million input tokens and $1.50/million output tokens.

(I was a little surprised not to see a price decrease there to better compete with the less expensive Claude 3 Haiku.)

The price drop is particularly notable because OpenAI are promising to make this model available to free ChatGPT users as well - the first time they've directly name their "best" model available to non-paying customers.

Tucked away right at the end of the post:

We plan to launch support for GPT-4o's new audio and video capabilities to a small group of trusted partners in the API in the coming weeks.

I'm looking forward to learning more about these video capabilities, which were hinted at by some of the live demos in this morning's presentation.

# 13th May 2024, 7:09 pm / generative-ai, openai, gpt-4, ai, llms, vision-llms, multi-modal-output