Moondream 3 vs LFM2.5-VL 3B: One Points At The Answer, One Just Answers

The most useful difference between M​oondream 3 Preview and LFM2.5-VL-3B is not in their benchmark tables. It is that M​oondream 3 Preview reasons about an image by pointing into it — its reasoning traces are grounded, referencing specific regions — while LFM2.5-VL-3B is explicitly a non-reasoning model that answers directly, a choice Liquid AI made to keep latency low for on-device work. Same image in, two entirely different things come out, and which one you want depends on whether a downstream system has to trust the answer or merely consume it.

Everything else follows from that fork. M​oondream 3 Preview is a sparse mixture-of-experts model at 9B total parameters with 2B active, released on 18 September 2025 by M​oondream, which was acquired by Prime Intellect in October 2026. LFM2.5-VL-3B is a dense 3.1B model — a 2.6B language backbone with a 400M SigLIP2 NaFlex vision encoder — released by Liquid AI on 12 August 2026 as a direct successor to LFM2-VL-3B. One is a specialist that was built to detect and point and can also answer questions. The other is a generalist that was built to answer and recently learned to point.

The metric collision you have to notice first

Put the two grounding numbers side by side and the comparison looks like a near-tie. It is not, because they are not the same measurement.

M​oondream's launch chart lists M​oondream 3 Preview at 91.1 on RefCOCO. Liquid AI reports LFM2.5-VL-3B at RefCOCO precision@1 of 87.9. RefCOCO accuracy and RefCOCO precision@1 are different statistics computed over the same dataset, and a 3.2-point gap between them tells you nothing about which model localises better. Anyone stacking those two figures in a bar chart is comparing a number to a differently-defined number.

• M​oondream 3 Preview grounding — RefCOCO 91.1, on a chart where frontier baselines (GPT-5, Gemini 2.5-Flash, C​laude 4 S​onnet) were queried through a template because they have no native grounding, and where some preview scores came from a 100-question sample rather than the full set. Vendor-reported and, as M​oondream says itself, not a fair fight.

• LFM2.5-VL-3B grounding — RefCOCO precision@1 87.9, up from 57.1 on the previous LFM2-VL-3B, which Liquid attributes to scaling synthetic grounding data. Vendor-reported, same-harness comparison against its own predecessor, which is the more meaningful axis.

• Detection and pointing — M​oondream 3 Preview: native detect and point skills, structured coordinate output, part of the architecture. LFM2.5-VL-3B: grounding added by training, returned as coordinates the same way, with function calling also new to the VL line (ToolSandbox 26.4 to 59.5, BFCL v4 20.5 to 32.5).

The honest read: Liquid AI has a clean before-and-after on identical harnesses and M​oondream has a cross-model chart with template caveats. Neither gives you a head-to-head. If grounding accuracy is the deciding factor in your evaluation, your own labelled set is the only instrument that will settle it.

What grounded reasoning costs you

M​oondream's query skill runs in reasoning mode by default. The model works through the question and its trace references the regions of the image it is using, which you can inspect in the playground by hovering over the text. For anything where a human has to audit the answer — a defect call on a production line, a security incident, a medical triage flag — that property is worth a lot. It is the difference between a model that says "no" and a model that says "no, here." p>The cost is tokens. Reasoning mode generates a trace before the answer, and M​oondream's own documentation offers the escape hatch: pass reasoning=False for simple questions, because the company knows the mode is not free. That is a real architectural admission — a well-designed model would not need a flag to switch the thought off.

LFM2.5-VL-3B has no equivalent. It was designed as a non-reasoning model, and Liquid AI is direct about the trade: answering directly keeps latency low for real-time and on-device use. There is no trace to inspect and no mode to disable. On a checkout line reading barcodes, or a phone translating a menu, that is the right engineering call and the missing trace is not a defect.

So the question is not which is better. It is whether anything downstream of your model reads the reasoning. If a human does, M​oondream's grounded traces are the feature you are buying. If only a program does, you are paying for tokens nobody will read.

Where 3.1B actually wins

LFM2.5-VL-3B's size claim is not marketing. It runs in under 3.3 GB of memory at 228 tokens per second on an Apple M5 Max, and 116 tokens per second on an AMD Ryzen AI Max+ 395, with GGUF, ONNX and MLX exports published alongside the native checkpoint. That is a model you can put on a laptop, and Liquid AI also ships it as a WebGPU Hugging Face Space that runs in a browser tab.

M​oondream 3 Preview has a smaller active parameter count — 2B against 3.1B — but the comparison is not symmetric, because a MoE checkpoint has to be fully resident even though only a fraction of it fires. The 9B of weights need somewhere to live. M​oondream's own answer to that is Photon, its local inference engine, which runs the model on N​VIDIA GPUs from Ampere onward and on Apple Silicon Macs. Different constraint: LFM2.5-VL-3B is small enough to fit anywhere, M​oondream 3 Preview is fast in the compute sense but large in the memory sense.

• Context — both 32K tokens. Genuine convergence, and worth noting given the two-year jump from the 2K windows of the previous M​oondream generation.

• Languages — LFM2.5-VL-3B lists sixteen: English, Arabic, Chinese, French, German, Italian, Japanese, Korean, Portuguese, Spanish, Vietnamese, Thai, Indonesian, Hindi, Russian, Polish, and doubled its vocabulary to 128,000 to support non-Latin scripts. M​oondream 3 Preview's card leads on visual capability rather than a language list; treat multilingual VQA as something to test rather than something promised.

• Document reading — LFM2.5-VL-3B: DocVQA 91.1, ChartQA 81.3, OCRBench v1 84.2, OCRBench v2 English 47.5, MMMU 48.4. M​oondream 3 Preview: DocVQA 88.3, ChartQA 86.6, POPE 89.0. Different harnesses, vendor-reported on both sides.

• What LFM2.5-VL-3B tells you not to do — Liquid AI's card says plainly it is not recommended for long-context or reasoning-intensive tasks such as visual web design or technical questions about blueprints. That sentence is more useful than most benchmark tables.

Deployment, and where a router fits

Neither of these models is served through OrcaRouter, and it is worth saying so directly rather than dancing around it: both are open-weight checkpoints you run yourself, and M​oondream 3 Preview's hosted endpoint — M​oondream Cloud — is being shut down on 13 November 2026 following the Prime Intellect acquisition, with unspent credits refunded.

What that changes is the shape of the decision. A team standardising on a self-hosted VLM usually ends up with a second, hosted model for the cases the local one handles badly, and at that point the integration work is the problem: two SDKs, two auth schemes, two retry policies, and a vendor outage that becomes your outage.

This is the case OrcaRouter's routing DSL is built for. You compose several models into a single call — the 3.1B local model answers what it can, a hosted model takes the residue — behind one key for more than 200 models, with provider list price passed through at zero markup and automatic failover when an upstream degrades. Since neither M​oondream 3 Preview nor LFM2.5-VL-3B is in that catalogue, the routing layer sits above them rather than replacing them: it is how you keep the cheap local path cheap without making a vendor outage your outage.

The licensing fork, which is sharper than the benchmarks

Two different licensing philosophies, and the difference matters more for a product decision than a 3-point score gap.

M​oondream 3 Preview ships under the Business Source License 1.1 with an Additional Use Grant (No Third-Party Service). Internal and production use, personal projects, research, benchmarks, fine-tunes, merges, quantisations, weight deltas, and use inside a product that does not substantially overlap M​oondream's paid offerings are all permitted. Selling a hosted API that exposes similar vision capabilities, managed "M​oondream-as-a-service", or embedding the weights in a paid SDK that delivers comparable capabilities requires a separate agreement. M​oondream 3.1 moved to the M​oondream Model License 1.0, which states the same rule more cleanly: component of your product is allowed, access to M​oondream itself as the product is not. Edge devices, customer-controlled VPCs, and running the model on leased cloud or neocloud GPUs are all explicitly on the allowed side.

LFM2.5-VL-3B is under Liquid AI's own LFM Open License v1.0. Read it before you ship, as you would any bespoke licence, but the families are different in kind: M​oondream's restriction targets one specific business — reselling the model — while Liquid's is a broad open licence with its own terms.

If you are a domain-specific builder, both are fine and the licence question is a formality. If your business is anywhere near "we host vision models for other people," M​oondream's licence is the constraint you read first and LFM's is the one you probably clear.

Which one to pick

Choose M​oondream 3 Preview if the answer has to be inspectable. If a human reviews the output, if the trace needs to show which part of the image drove the decision, if detection and pointing are the primary product rather than a feature — that grounded reasoning is the whole reason to pay 9B of memory for a 2B-active model. Budget for Photon or a GPU, and plan around the hosted endpoint closing.

Choose LFM2.5-VL-3B if the model has to sit on the device. If you are shipping to a phone, a laptop, a factory floor box, or a browser tab, and the workload is single-turn, high-throughput and low-latency — reading screens, transcribing documents, detecting objects in a stream — then a dense 3.1B model that answers without thinking is exactly what you want, and its 228 tokens per second on an M5 Max is the number that decides it.

Pick both if you have the budget for it, and route between them: LFM2.5-VL-3B on every frame because it is nearly free to run, M​oondream 3 Preview on the frames where a grounded answer is worth waiting for.

What the numbers cannot tell you is whether any of it has gone wrong in the field: the documented AI incidents and agent failures around models in this class are collected in the OrcaRouter AI Incident Archive, where each entry carries its own sources, severity, confidence grade and disputed flag.