Introducing Clef-omni with full multimodality, plus a faster Clef and a cheaper Clef-flash: Introducing Clef-omni with full multimodality, plus a faster Clef and a cheaper Clef-flash We are expanding the Clef decision model family with Clef-omni, natively processing audio, video, images, and text in a single pipeline. We’ve also lowered Clef-flash pricing and boosted Clef inference speeds by up to 2.0x.

2026-10-10T06:13:56Z

We are expanding the Clef decision model family with Clef-omni, natively processing audio, video, images, and text in a single pipeline.

Introducing Clef-omni with full multimodality, plus a faster Clef and a cheaper Clef-flash We are expanding the Clef decision model family with Clef-omni, natively processing audio, video, images, and text in a single pipeline. We’ve also lowered Clef-flash pricing and boosted Clef inference speeds by up to 2.0x.: Research context

This is just the beginning, and we’ll continue to get better, faster, cheaper, and more innovative. 7× One of the optimizations we made was to move to SGLang for serving the model. No new model weights are released because most of the optimizations we made are at the serving infrastructure layer, not materially to the model weights or architecture. We also released the model open-weights on HuggingFace. How we trained Clef In the same week that Jev came out, we posted about some experiments we had with our own homegrown decision model. 8 Invoice processing Primary action 82. We are expanding the Clef decision model family with Clef-omni, natively processing audio, video, images, and text in a single pipeline. We built this on a Qwen3-Omni-30B-A3B-Instruct mixture-of-experts (MoE) foundation, where the model was already capable of processing text, imagery, audio, and video directly within a single pipeline. Clef-omni takes audio, video, image, and text input Our new Clef-omni model is able to take audio, video, image, and text input. With Clef, we supported images and video frame arrays, but Clef-omni is able to take in audio (wav or mp3) and video (mp4 or webm) alongside text and images. Today, we’re releasing Clef-omni, which takes in audio and video input alongside text and image. Check out our developer docs for the new Clef-omni model. Check out the developer docs for the most up-to-date pricing information at all times, including more detail on how image and audio modalities are converted to input tokens for pricing. At Cloudflare, we’ve had teams experimenting with Clef as a model that can help classify over general domains. Specifically at Cloudflare, we’ve been testing our new Clef model on our Threat Intelligence team to help us classify website domains. Check out our new speeds below: Input size Before: median / p95 (ms) Now: median / p95 (ms) Median speedup ~800 tokens 262 / 438 152 / 351 1. And our threat intelligence team is using Clef to help detect malicious domains. As a user, you can imagine how a 2x savings in latency and results can help us improve our threat intelligence workflows and be faster in identifying malicious or legitimate domains. We’ve made a few optimizations and can now offer Clef-flash at cheaper prices. We also cut the price of Clef-flash so it is now cheaper than Jev, and we made Clef faster. 6 Clef-flash model is now cheaper We heard your feedback — you want a decision model affordable enough to incorporate into any workflow. Clef model is now faster Speeds are also faster for the Clef model, with new optimizations to the hosted version of Clef on Workers AI. For a quick start, here’s how you can send new modality inputs to Clef-omni: Clef-omni expands our open-weight decision architecture to natively handle multimodal workflows. Workflow Metric Clef-Omni Clef Clef-flash Jev Invoice processing Exact actions 60. We were able to optimize the model to the point where Clef-flash is now cheaper than Jev while still being performant and high-quality. 15 per M input tokens However, one trade-off we have to make in order to have cheaper pricing is to cut the context window of Clef-flash. We are introducing Clef and Clef-flash, open-source decision models hosted on Workers AI for high-speed classification and agentic workflows. Following last week’s release of Clef and Clef-flash, Cloudflare’s open-weight decision models, we decided to bring forth more gifts. Benchmark Clef-omni Clef Clef-flash Jev BFCL · case exact 98. Our Clef model remains at a 64k context window, and we encourage folks with larger context needs to switch to Clef instead of Clef-flash. The coolest thing about watching teams adopt Clef is that the capability to do detections and classifications have moved into the model layer, where it is accessible for anyone to start incorporating decision models into their use cases. Today, we’re releasing two Cloudflare-trained decision models, Clef and Clef-flash, hosted on Workers AI. Clef is currently the leader when evaluated against the Jev Decision Index, you can view full results on the live benchmark demo site. Check out the table below for benchmarks, or view the scores on our live decision index demo site: Benchmark Clef Clef-flash Jev DiffusionGemma Jev Kev 9B Laya BFCL · case exact 98. Notably, our Clef-flash performs exceptionally well, given how much faster it is. Workflow Clef Clef-flash Jev Invoice processing 64. 6 Across the 43 eval benchmarks that we ran, our Clef models beat the decision models on latency (except for Laya which is very fast but trades off quality in the benchmarks above): Benchmark Clef Clef-flash Jev DiffusionGemma Jev Kev-9B Laya Median latency · ms 209. The larger Clef model is your more powerful precision model, while the Clef-Flash model is great for latency-critical decisions. If you’d like help tuning Clef for a specific workload, we are also offering fine-tuning services — first as a hands-on partner with our forward-deployed engineer (FDE) team, and then later as a self-serve fine-tuning platform for customers to train and redeploy the model onto Cloudflare. Rather than generating intermediate text to produce structured answers, Clef and Clef-flash derive schema choices directly from internal backbone representations. Because Cloudflare has more than 15 years of network data across different domains, we can fine-tune a model to fit these specific use cases which is more accurate and faster than our generic Clef

Key findings

model. 24 per M input tokens Clef-omni – launched today at $0. The hosted version of Clef-flash now has a context window of 24k, instead of 64k as previously advertised. Based on this data, we decided to make the call to drop the context window to 24k for Clef-flash in order to price it at a more accessible entry point. You can also self-host Clef since we released the weights, and the new SGLang launch commands are available on our updated Hugging Face repo as well as the SGLang cookbooks. Clef is fully Jev-API compatible and available via AI Gateway, so it is as simple as changing the model ID to try out Clef today. Also launching: a new reinforcement learning platform that allows developers to fine-tune decision models using their own data. This classification took our Clef model 2. Although the market is getting increasingly saturated with decision models, Clef has some unique properties that make us excited to release it to the public. Third, our model is accurate and powerful, scoring competitively against other decision models on the market across various quality benchmarks. 15 We also ran benchmarks across Typesafe’s own eval suite and our Clef models fared well, beating Jev in 3 out of 4 areas. 5 On top of the latency benefits from the model itself, our Clef models are hosted on Workers AI. Clef also produces strictly typed outputs similar to Jev and is fully API-compatible, so you can make the swap extremely easily. You can get started with the Clef models today, starting with our developer documentation or play around with the open-source model on the Hugging Face repo. How fine-tuning can extend the capabilities of Clef We heard a lot of internal use cases that required fine-tuning our Clef model to be built into our agentic workflows at Cloudflare. To do this, we leverage the primitives that we already have built on our Cloudflare platform: Cloudflare AI Gateway – pass all your AI traffic through AI Gateway and automatically create a dataset of requests for your use case Cloudflare Workers AI – generate rollouts against the base Clef model Cloudflare Containers – RL sandbox for scoring and replaying agent actions [NEW] Trainer – update weights of fine-tuned Clef model Cloudflare Workers AI + BYO Model – redeploy the fine-tuned model on Workers AI This combines a few work-in-progress pieces of the AI Platform that we’ve been working on, including AI Gateway that captures your AI traffic so you can leverage your own request/response data, Containers for RL Sandboxes, and Workers AI’s Bring Your Own Model (Cog) work that has been progressing since our acquisition of Replicate. 038 per M input tokens Clef – remains at $0. The model weights on Hugging Face are untouched and were trained to support a 256k context window should you choose to self-host it. These models are smarter, faster, and fully Jev-API compatible, so you can experiment with these hosted models easily. Lastly, we’re excited to debut our new reinforcement learning (RL) product, which allows customers to fine-tune Clef to suit their use cases as well. Secondly, our model has a 64k context window (compared to Jev’s 32k), which allows users to squeeze more input state for the model to classify against. We shortlisted some evaluations below that are important for decision-making as defined by the Jev Decision Index and scored some of the more popular models on the market for it. When you fine-tune a model, you may give up some general purpose performance in exchange for higher accuracy in a specific domain. Our new RL service We are offering a service to help customers fine-tune Clef to suit their workloads with our hands-on FDE team. From that, we’ll learn from our hands-on experiences to build a self-serve platform that customers can use to capture data, fine-tune, and redeploy the model, all on Cloudflare. Try out the Clef models hosted on Workers AI, download the weights on Hugging Face if you’d like to explore for yourself, and reach out if you have fine-tuning use cases you’d like us to help with. How is Clef different from other decision models? We chose Clef as the name of our family of decision models, as it serves similar purposes, and the CF hearkens to Cloudflare. We’re excited to bring you more from the Clef family of models, including new modalities like audio and video, as well as making the model more accessible with cheaper pricing and faster speeds. For today, we’re excited to keep up the momentum with new additions and improvements to our Clef family of models. Try out our Clef family of models, and tell us what you think. Try it out today We’re excited to launch our first Cloudflare-trained ML model from the Workers AI team today. Calibrated, schema-bound decisions now seamlessly handle text, images, audio, and synchronized video in a single API call. 5 seconds, all in a single API call. We’ve also lowered Clef-flash pricing and boosted Clef inference speeds by up to 2. This changes the paradigm for decision models, which have been largely text-only since the debut of Jev from TypeSafe. This is different from Jev, which only does text classification today. Over the last few weeks, there has been lots of buzz around decision models such as Typesafe AI’s Jev System One model. This

Operational implications

brings several core upgrades to the Clef model family. We want Clef to be able to make decisions that scale for you across all your agentic workflows, and now our pricing also incentivizes that. This means that you could put Clef into the hot path for agents to make decisions and combine that with one of our LLMs on Workers AI to take action. How are people using Clef? 8-27B for Clef and Qwen3. Clef builds upon this concept, but uses a different base model as the backbone. A few highlights of how teams at Cloudflare are using Clef: Our public GitHub docs repo uses Clef to instantly detect and close spam issues. While classifier models have been around for some time, Jev introduces a new decision model concept into the world of AI — a model that produces bounded structured outputs cheaply, quickly and consistently that can be added into a workflow when a decision is required. We’re working with internal teams already to figure out how we can post-train Clef to create powerful ML models that boost our impact and improve workflows across Cloudflare. We currently use Qwen as the base model and post-trained it to suit decision model use cases. What is a decision model? A decision model makes classifications to help agents decide how to act, based on certain probabilities. Instead of setting up cascading pipelines of models that transcribe speech-to-text, or splitting audio and image channels from video, you can just call one model to make decisions across any modality. In fact, the story of Clef came together over the course of less than a week. Because they are hosted on Cloudflare’s infrastructure, we’re able to take advantage of our GPUs at the edge, leading to low network latency and faster decisions. We decided we wanted to do something in the decision model space on a Friday evening, trained the model over the weekend, and launched it on Thursday. Because we skip output token generation since Clef models are not Large Language Models (LLMs), we remove the overhead of transcribing or captioning incoming files. This contrasts with the world of Large Language Models (LLMs), which are largely non-deterministic, but are open-ended enough to reason and generate text and tool calls for agentic workloads. This means that we were able to achieve a few novel things with Clef: we improved accuracy of the model in classification, constrained it to output only probabilities instead of text generation, and made it faster than Jev and the base Qwen models. 55 We also benchmarked against the TypeSafe evals to show how Clef-omni performs. Text-only decisions return in about 130 ms at the median, image inputs in about 150 ms, and audio clips in a few hundred milliseconds. These models are capable enough to work over any set of inputs without constantly retraining the model to incorporate new classification categories. The decision step is non-autoregressive, so there’s no intermediate text to generate token by token, making Clef significantly faster than autoregressive LLMs. Although the model game is still early for Cloudflare, innovation and iteration is in our DNA, and we apply these principles to everything we do. First, it has a vision encoder so it’s able to take in images and classify visual content. For example, internal teams want a classifier model to be able to evaluate Trust & Safety submissions, help us triage Cloudflare Support requests, or even to be built-in to our Bot products to decide if a crawler is a good bot or bad bot. A decision model is analogous to a music clef because it helps define the domain of the context and the subsequent notes (actions) that follow it. We train the model with the same techniques as we did with Clef — freezing the Qwen3 backbone, training low-rank adapters (LoRA), and applying our standard post-training approach: combining label-smoothed cross-entropy loss with Brier score calibration. 5-9B for Clef-flash, we jointly optimized the routing head alongside rank-256 low-rank adapters. Now, all you need to do is call a model to instantly parse through input data with precision. We are now one step closer to a model that is able to interact with the world as we experience it — through audio, visual, and textual communication, all in one. We believe that Clef has the ability to disrupt the way we use agents, which fits naturally into Cloudflare’s mission of being the agent cloud. Clef-omni then pulls candidate values directly from internal embeddings using our two-stage attention routing: every valid option gathers key evidence from the input (regardless if they are buried in text snippets, visual elements, or audio streams) before field vectors cross-attend across the full context to compute confidence scores. During inference, Clef uses Qwen for a prefill-only pass, then scores the valid schema choices in parallel. In production, Clef-omni executes a quick prefill pass across the complete payload, scoring all modalities and valid parameter options simultaneously. We worked with the SGLang team and made pull requests to incorporate Clef (PR #42721), which will be released in SGLang 0. Media elements map straight into the unified sequence, where video and audio are synced with visual frames for joint processing. The interest in Jev shows the need for a fast, small, specific, classifier model, and we chose this to be our niche to start experimenting with RL environments.

Try out the Clef models hosted on Workers AI, download the weights on Hugging Face if you’d like to explore for yourself, and reach out if you have fine-tuning use cases you’d like us to help with.

Read the original sources cited in the article for the full material.

Frequently asked questions

What does the collected research say about Introducing Clef-omni with full multimodality, plus a faster Clef and a cheaper Clef-flash We are expanding the Clef decision model family with Clef-omni, natively processing audio, video, images, and text in a single pipeline. We’ve also lowered Clef-flash pricing and boosted Clef inference speeds by up to 2.0x.?

We are expanding the Clef decision model family with Clef-omni, natively processing audio, video, images, and text in a single pipeline.

What are the key details from Cloudflare Blog?

We are introducing Clef and Clef-flash, open-source decision models hosted on Workers AI for high-speed classification and agentic workflows.