Liquid AI’s d1 is a fit for tasks with a defined answer set—such as yes/no classification, choosing among known options, or assigning a score—because it returns probabilities without generating output tokens. A vision-language model (VLM) is usually the better fit when a task needs open-ended image interpretation or a natural-language answer. Choose by testing the actual task, input modality, required output, quality, latency, and deployment constraints—not by “zero tokens” alone.
What does Liquid AI d1 return?
d1 is a decision model: it evaluates a situation and returns probabilities for a bounded set of possible answers in a single forward pass. Liquid AI describes the distinction this way: “Decision models answer questions about a situation with a probability for each possible answer.”
Liquid’s documentation defines three question types:
- Noul: a yes/no answer expressed as a probability between 0 and 1, such as “Is this message spam?”
- Choice: a probability distribution over named alternatives, such as which department should handle a support ticket.
- Score: a probability-weighted position on an ordered rubric, such as how urgent an issue is.
An application can request multiple decisions about the same state. The result is structured for software to consume directly: for example, to route a request, apply a threshold, rank options, or select an action. That differs from asking a model to write an explanation for a person. Liquid AI’s d1 launch article and its decision-model documentation describe this interface.
Free tools Windows power users keep installed
One-click scans. No signup required.
#1 Best Overall
- EVOLUTION AMD RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
How is d1 different from a vision-language model?
“Vision-language model” describes a broader class of models that take visual and text inputs and may generate richer interpretations or text. d1’s defining feature is its constrained decision interface: probabilities over fixed outcomes, rather than generated output tokens. Liquid AI separates Decision Models from Vision-Language Models in its model catalog.
The labels do not mean the models have wholly separate foundations. Liquid says d1-3B is trained from LFM2.5-VL-3B, while d1-omni-600M derives from an encoder backbone with vision and audio components. The practical question is what the application must return, not simply whether it processes an image.
- Test d1 when the valid answers are known in advance and the application needs a probability, a choice, or an ordered score.
- Consider a generative VLM when users need a description, explanation, summary, or flexible answer.
- Consider a two-model workflow when most cases are bounded decisions but uncertain or open-ended cases need generated interpretation.
When are zero-output-token decisions useful?
d1 is most relevant when a system already knows its possible outcomes and needs a structured result to trigger the next step. Liquid AI’s demonstrations illustrate several candidate workflows; they are examples to evaluate, not guarantees of production performance.
Classification and filtering
Liquid demonstrates filtering support tickets and describes yes/no classification, including identifying cancellation intent. Its Smart Filter example uses 150 tickets and compares the filter with hand labels. A production team should still measure its own false positives and false negatives on representative traffic.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Routing, filing, and retrieval support
Liquid shows documents being filed into folders and subfolders. A Choice question can also select a support department. Another demonstration uses d1 to locate relevant code in a repository and organize search questions into folders; this is one possible workflow, not evidence that d1 replaces general code-search systems.
Rank #2
- Built for Local AI Development: AMD Ryzen AI Halo is designed for local AI development and inference, featuring 128GB unified memory and support for up to 200B parameter models to build and run intensive AI workloads locally.
- 128GB Unified Memory: Features 128GB LPDDR5x unified memory at 8000 MT/s with 256 GB/s memory bandwidth, providing a shared memory pool across the CPU, GPU, and NPU to support larger AI models.
- AMD Ryzen AI Max+ 395 Processor: Features 16 cores, 32 threads, and Zen 5 architecture, paired with AMD Radeon 8060S integrated graphics featuring 40 RDNA 3.5 compute units and an AMD XDNA 2 NPU with up to 50 TOPS.
- Linux AI Developer Platform: Purpose-built for Linux-based AI development with full AMD ROCm software support and preloaded tools, models, and workflows optimized for local AI development.
- Compact, Connected Design: Includes a 2TB M.2 SSD, 10GbE LAN, Wi-Fi 7, Bluetooth 5.4, USB-C connectivity, and HDMI 2.1b.
Choosing an action
A web-agent demonstration has d1 choose the next available action on a flight-search website. This bounded action-selection setup is a natural candidate when the interface exposes a finite menu of valid next steps.
Visual inspection and screenshots
Liquid reports 85–97% accuracy across four VisA tasks sorting good and defective circuit boards, candles, cashews, and chewing gum. The company says d1 was not trained specifically for those inspection tasks. These figures describe Liquid’s reported tests; they do not establish accuracy for another factory, defect type, camera, or dataset.
Liquid also reports demonstrations involving interactive screenshots: adding a Tetris screen raised d1’s score from 70 to 81 lines cleared, and it solved 12 of 12 Wordle games in an average of 3.8 guesses using screenshots. These are demonstrations, not independent benchmarks.
Recommended Free Tools
Selecting useful context
In a coding-agent context-compaction demonstration, Liquid reports removing 52% of tokens while retaining the outputs needed for the next task. That result applies to the sessions and setup Liquid describes, not to every agent or workload.
For consequential automated decisions, validate on representative data, choose thresholds based on error costs, log error types, and define fallback behavior. Keep human review where the cost of a mistaken decision warrants it. Liquid’s cited materials do not establish a universal risk threshold or a general production-accuracy guarantee.
Rank #3
- EVOLUTION RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
Which d1 models and input modalities are available?
Liquid AI’s October 7, 2026 release names two open-weight models:
- d1-3B: based on LFM2.5-VL-3B; accepts text and images.
- d1-omni-600M: an experimental checkpoint based on LFM2.5-Encoder-350M; accepts text plus image or text plus audio. Liquid says it remains under active development.
Liquid says both are available on Hugging Face and have day-one llama.cpp support. Its October 5 announcement also describes a hosted d1 model through Liquid AI’s API. Model versions, service access, and third-party support can change; check the current open-model release and API announcement before implementation.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsWhat do the benchmark results establish?
Liquid reports a score of 48.57 for d1-3B on Decision Index v0.2.1 public split. In its October 7, 2026 release, the company says this is ahead of every model under 10 billion parameters and on par with Decider 35B-A3B on that benchmark. These are vendor-reported results, not an independent evaluation.
Liquid’s table of seven public text benchmarks reports means of 82.9 for d1-3B, 78.4 for d1-omni-600M, 81.1 for Decider 4B, and 77.1 for Decider 2B. The listed benchmarks are SQuAD 2.0, Civil Comments, MASSIVE intent, PubMedQA, BoolQ, XNLI, and PAWS-X. A mean across different tasks can conceal a weakness on an individual task and does not predict performance on a particular production workload.
A separate Liquid launch comparison says d1 matched or beat GPT-6.1 Sol on four of six applications, cost 19 to 200 times less, and answered faster on every task. Liquid says it ran each application once on October 5, 2026, using its d1 Playground comparison script. The chat models received one chat message and JSON output at default reasoning settings; costs used list prices without prompt-cache discounts, with d1 calculated at $0.04 per million input tokens. Smart Filter used 150 tickets, Smart Folders used 105 passages, and several code and compaction questions were written after d1’s pipeline was set. Treat this as a vendor-reported snapshot with that method, not a general price or quality guarantee.
For vision, Liquid says d1-3B retained the vision capabilities of its LFM2.5-VL-3B backbone on standard vision benchmarks, but the October 7 release does not report the private vision split. Liquid also says dedicated audio decision benchmarks remain an open problem. The published public text scores therefore do not establish general vision or audio superiority. See the release and benchmark details.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteHow fast is d1 on edge hardware?
Liquid’s October 7, 2026 release reports the following single-question d1-3B latencies:
| Device | Reported latency |
|---|---|
| Jetson Orin Nano | 50 ms (Liquid AI, 2026) |
| Jetson AGX Thor | 16 ms (Liquid AI, 2026) |
| Jetson AGX Orin 64 GB | 26 ms (Liquid AI, 2026) |
| Apple M5 Pro | 30 ms (Liquid AI, 2026) |
| NVIDIA RTX 4090 | 8 ms (Liquid AI, 2026) |
These are selected measurements, not a universal latency promise. The same release reports 1,640 ms on Jetson Orin Nano for a 3.4K-token state and 202 ms for a 384-pixel image. It measures one request at a time and includes other state and image cases. Compare candidates using the same hardware class, state length, image resolution, number of questions, runtime, quantization, batch shape, and warm or cold conditions. Liquid also demonstrates d1-3B served on Jetson hardware in an Isaac Sim setup with NVIDIA collaboration.
How do hosted API and open-weight deployment differ?
Liquid’s October 5 API announcement says billing is based on input tokens, with no output tokens. Under the pricing method stated there, images count as 1.5 tokens per 32×32-pixel patch; a 1024×1024 image therefore counts as 1,536 input tokens. The same announcement says Vercel and OpenRouter were text-only at publication, with vision planned later. These service details are time-sensitive; verify current terms and modality support with the API announcement and providers.
Hosted API access and open-weight deployment are different operational choices. For either route, assess the actual data path, privacy requirements, hardware or service costs, and integration needs; an on-device option does not by itself establish that an application meets its privacy or operational requirements.
How should you compare d1 with a VLM for your workload?
Run both candidates on the same representative inputs and judge them against the application’s requirements. A useful evaluation covers:
| Axis | Question to answer |
|---|---|
| Output shape | Are valid answers yes/no, a named option, or an ordered score, or must the system produce arbitrary text? |
| Input modality | Does the task use text, images, or audio, and does each candidate support that exact combination in the chosen deployment? |
| Task quality | On representative labeled examples, what errors occur? Are probabilities calibrated, and how does threshold choice affect outcomes? |
| Latency | What is end-to-end latency at the real state length, image size, batch size, runtime, and target device? |
| Integration | Can the application consume a probability distribution, or does it need generated explanations, tool use, or conversational turns? |
| Cost and privacy | What are the current API or hardware costs, and where does the data travel? |
A two-stage design is worth testing when a cheap, high-volume bounded decision can handle routine cases and uncertain or open-ended cases can be routed to a VLM. This is an architectural option inferred from the models’ output shapes, not a performance result established by Liquid’s cited materials.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
