LFM3-VL-2B vs Olmo 3 7B Think

LFM3-VL-2B 90

Chat template 20, chat_template.jinja in the repository (12,129 characters), with roles, images and video, tools, deferred tools, tool results and the thinking switch. Tool-calling format 25, the card and Liquid's tool-use page document tools passed through tools= in apply_chat_template, calls returned as XML and results sent back as tool messages. Trained tool calling 20, the template loops over every entry in an assistant turn's tool_calls, so several calls fit in one turn, and the card says its fine-tuning mix included tool calling and agentic traces. Structured output 5, the card's vLLM and SGLang commands set a structured-output backend and its fine-tuning table has a JSON extraction notebook, but nothing says how to ask this model for JSON. Liquid's API guide documents response_format with lfm3-1.7b as its example. Reasoning control 10, Liquid's chat-template page documents enable_thinking=False in apply_chat_template for LFM3, with reasoning on by default inside think tags, and the card documents preserve_thinking and reports a no-thinking configuration in its benchmarks. Context 10, 128K tokens of input and output per the card, and max_position_embeddings 128000 for the language model in config.json. 20 + 25 + 20 + 5 + 10 + 10 = 90.

Olmo 3 7B Think 35

The chat template earns 20 for chat_template.jinja in the repository. The tool-calling format earns 10, since only the template and vLLM's own parser describe it. Trained tool calling earns nothing, because Ai2's report says tool calling was trained into the Instruct model, not Think. Structured output earns nothing. Reasoning control earns nothing, since it always thinks. Context earns 5 for 65,536 tokens trained. 20 + 10 + 0 + 0 + 0 + 5 = 35.

LFM3-VL-2B 40

File safety 15, safetensors only, no .bin, .pt or .pth files (25). Provenance 15, files in the LiquidAI organisation, which Liquid's docs page for the model and its blog post link to (10). Safety evaluation 0, no safety results on the card or in Liquid's blog post for the LFM3 family, neither of which mentions safety testing. Agent guidance 0, Nothing on prompt injection, an instruction hierarchy, what refusals cover or safeguards to add. Licence risk 10, 15 less 5 for the $10M annual revenue threshold on commercial use. Security programme 0, no way to report a flaw found. 15 + 15 + 0 + 0 + 10 + 0 = 40.

Olmo 3 7B Think 75

File safety earns 25, safetensors only and nothing from the repository runs. Provenance earns 15, files in the allenai organisation Ai2 links to, with a checksum for every file. Safety evaluation earns 20, as Ai2 publishes its safety results and the OLMES harness to rerun them. Agent guidance earns 5 for the card's general cautions. Licence risk earns 5, losing 5 for disallowed uses beyond the law and 5 because Ai2 can update the guidelines and end access at its discretion. The security programme earns 5 for a reporting route, a security.txt now past its expiry and a channel for security concerns in the guidelines, with no public advisories. 25 + 15 + 20 + 5 + 5 + 5 = 75.

LFM3-VL-2B 77

Model card 17, Intended use (questions about video, documents, charts and app screens, finding objects, spatial reasoning), a limitation (not recommended for coding, multi-turn conversations or complex reasoning), architecture (language model, 14-layer vision encoder and projector), parameters (1.92B), context (128K) and 16 languages. Training data 5, A general description: the base's 22T tokens of pre-training with the kinds of data added since LFM2, then vision mid-training on image and video data, supervised fine-tuning, vision reinforcement learning and distillation. Last change 20, the last commit, on 9 November 2026, added the OpenVINO checkpoint to the card's inference table, a day before this check. Responsiveness 20, Of the 5 discussions that aren't pull requests (27 to 31 October 2026), 3 have a reply from a LiquidAI member, which is most. Version policy 5, LFM3 came out in new repositories and the LFM2.5 models stay in theirs. File health 10, Config, template and card agree on the lfm3_vl architecture, the 128,000-token context and the vision settings in processor_config.json. 17 + 5 + 20 + 20 + 5 + 10 = 77.

Olmo 3 7B Think 52

The model card earns 17, missing only the context length. Training data earns the full 20, as every dataset is released with sources and sizes. The last change earns 10, a card fix on 25 June 2026, within 180 days. Responsiveness earns nothing, with no Ai2 reply among the 4 replies in the discussions. Version policy earns 5, with earlier and later models in their own repositories and no dated notes. File health earns nothing, because the template's default prompt contradicts the card's on the cut-off and a stray swap file and an unanswered report sit in the repository. 17 + 20 + 10 + 0 + 5 + 0 = 52.

LFM3-VL-2B 85

Memory 30, Liquid's Q4_K_M GGUF is 1.10 GB with its 0.20 GB projector file, and 1.30 GB × 1.2 = 1.6 GB, which fits 8 GB, well inside the 16 GB line. Runtimes 15, Released llama.cpp has the lfm3 architecture and the LFM3-VL projector (src/models/lfm3.cpp and tools/mtmd at b12242), Ollama's library has an lfm3-vl entry, and LM Studio's catalogue lists liquid/lfm3-vl-2b. Builds 15, Liquid's GGUF repository holds Q4_0, Q4_K_M, Q8_0, F16 and BF16 files, and Liquid publishes 4-bit and 8-bit MLX builds and an ONNX repository, all in the LiquidAI organisation. Terms 10, the LFM Open License v1.0, a custom licence that lets anyone use it commercially for free below a threshold of $10M annual revenue. Download 10, the Hub API reports gated false for the model and its GGUF, MLX and ONNX repositories on 10 November 2026, so the files download with no account and nothing to accept. Hosted 5, Liquid's own API lists lfm3-vl-2b among its models, so the publisher hosts it. 30 + 15 + 15 + 10 + 10 + 5 = 85.

Olmo 3 7B Think 75

Memory earns the full 30, since Ollama's 4-bit build is 4.5 GB and fits 8 GB (per the 21 October check of Ollama's library). Runtimes earn 25, with released llama.cpp (b11963), Ollama (v0.40.14, pinning b11963), vLLM, LM Studio and Transformers all running it. Builds earn nothing, because Ai2 publishes no quantised build of its own. Terms earn 10, since the card binds Apache 2.0 to Ai2's Responsible Use Guidelines, a use policy. Download earns 10, with no gate. Hosted earns nothing, as the Hub's provider mapping is empty and Ai2 has no API for it. 30 + 25 + 0 + 10 + 10 + 0 = 75.