Devstral Small 2 vs Ling-3.0-tiny

Devstral Small 2 60

File safety 25, safetensors only, no .bin, .pt or .pth files and no Python in the repository. Provenance 15, files in Mistral's verified mistralai organisation, which Mistral's site links to per our Mistral Large 4 listing (10). Safety evaluation 0, no safety results on the card. Agent guidance 0, the card has no limitations or safety section, and nothing on prompt injection, instruction priority or what the model refuses. Licence risk 15, Apache 2.0 has no use policy, threshold or change clause. Security programme 5, mistral.ai's security.txt gives a security contact and a submission policy, as read for the Mistral Large 4 listing (5). 25 + 15 + 0 + 0 + 15 + 5 = 60.

Ling-3.0-tiny 50

File safety gets 15, safetensors and GGUF only, but the card's vLLM command runs the repository's config code with --trust-remote-code. Provenance gets 15 for files in inclusionAI's own organisation with a checksum on each. Safety evaluation is 0, since SingProbe's results grade the probe, not the model. Agent guidance gets 5 for the SingProbe guardrail inclusionAI publishes for this model, with no statement of trained refusals. MIT carries no licence deductions (15). No security contact or disclosure policy was found (0).

Devstral Small 2 46

Model card 11, Intended use, the architecture, the parameters and the context are on the card. Training data 0, Nothing on the card about the training data. Last change 10, the last commit, on 15 July 2026, updated the files to match mistral-common, 118 days before this check and within 180. Responsiveness 10, Of the 23 most recent discussions that aren't pull requests (9 December 2025 to 6 November 2026), 11 have a reply from a mistralai member, which is some. Version policy 5, Devstral Small 1.1 stays in its own repository, which this card links as the predecessor, with no dated note on the older card that we read. File health 10, Config, template, tokenizer and card agree, and the June and July commits fixed stray tokens and brought the files in line with mistral-common. 11 + 0 + 10 + 10 + 5 + 10 = 46.

Ling-3.0-tiny 56

Model card 11, intended use, architecture, parameters and context are there, limitations, languages and a cut-off aren't. Training data 15, a data card with sources, size, cut-off and filtering. Last change 15, the newest commit is still the card change of 24 September 2026, 41 days before 4 November. Responsiveness 10, six of 16 discussions answered by an inclusionAI member, some. Version policy 5, a first release at this size. File health 0, the card's MTP recipe contradicts the files and a broken index size waits without a reply.

Devstral Small 2 90

Memory 25, the smallest build at 4 bits from a runtime's own library is Ollama's devstral-small-2 at Q4_K_M, 15 GB, and 15 GB × 1.2 = 18 GB, within 24 GB. Mistral's own FP8 files are 25.79 GB. Runtimes 25, Released vLLM 0.33.0, SGLang 0.5.23 and Transformers 5.22.2 have the mistral3 model, llama.cpp b12242 has the architecture with a GGUF from its own organisation, and Ollama's library has devstral-small-2. Builds 10, one quantised precision from Mistral, the FP8 weights it releases. Terms 15, Apache 2.0 in the card's metadata and text, an OSI licence with nothing added. Download 10, the Hub API reports gated false on 10 November 2026, so the files download with no account and nothing to accept. Hosted 5, Mistral's own API serves Devstral Small 2: the card says the model is available through its API and points there when a local runtime underperforms. 25 + 25 + 10 + 15 + 10 + 5 = 90.

Ling-3.0-tiny 95

Memory earns the full 30, since inclusionAI's own Q4_K_M GGUF is 4.8 GB, about 6 GB with a fifth added. Runtimes earn the full 25, from released Transformers, vLLM, SGLang and llama.cpp, and Ollama, whose v0.40.14 (2 November 2026) pins llama.cpp b11963 with the bailingmoe3 architecture (Ollama v0.40.7 of 20 October already pinned b11738, which has it). LM Studio has nothing. Builds earn 15 for FP8, INT4 and four GGUF precisions from inclusionAI. MIT with nothing added gives terms the full 15 and the ungated repositories give download the full 10. Hosted too scores 0, because the Hub's model page lists no inference provider and no inclusionAI or Ant Group API serving it was found.

Devstral Small 2 65

Chat template 20, chat_template.jinja in the repository (11,362 characters), with the tokenizer's tekken file for mistral-common. Tool-calling format 25, the card documents tools in the OpenAI function format with an example, names the mistral tool-call parser with auto tool choice for vLLM, shows the same for SGLang, and ships test prompts with the expected tool-call output. Trained tool calling 10, the card says the model excels at using tools, and its tests cover one tool call and calls made one at a time in sequence. Structured output 0, Nothing on the card about JSON or schema-constrained answers for this model. Reasoning control 0, no reasoning mode is documented, and a June 2026 commit removed stray thinking tokens from the files. Context 10, 256K tokens per the card, served at 262,144 in its vLLM command. 20 + 25 + 10 + 0 + 0 + 10 = 65.

Ling-3.0-tiny 60

Chat template 20, a Jinja file in the repository. Tool-calling format 10, the format is only in the template and the card names the ling3 parser in commands without describing it, while SGLang uses another parser. Trained tool calling 20, several calls a turn and agentic results on the card. Structured output 0, not documented. Reasoning control 0, thinking turns off per request but the card doesn't say where the reasoning comes back. Context 10, 131,072 tokens native.