Devstral Small 2 vs GLM-5.3-Flash

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.

GLM-5.3-Flash 60

Memory earns nothing, since Z.ai's smallest build is the 328.4 GB FP8 one, about 394 GB with the fifth added, and neither Ollama nor LM Studio has a local build (both list it as cloud only). Four runtimes run it in a released version, llama.cpp, vLLM, SGLang and Transformers, for 20, while Ollama and LM Studio don't. Z.ai publishes one quantised build, FP8, for 10. The MIT licence with nothing added gives the full 15, the files download with no account or gate for 10, and Z.ai's own API plus six providers on Hugging Face serve it, for 5.

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.

GLM-5.3-Flash 58

The card covers architecture, parameters and languages and misses four of the seven items, for 8. Training data is 5, a 30T-token corpus described with no sources or cut-off. The last change, a chat template speed-up on 15 October in both repositories, is within 30 days on 2 November, for 20; Z.ai's GLM-5 GitHub repository has changed only its docs since. Responsiveness is 10, as zai-org answered 14 of the 38 recent discussions others opened (26 August to 30 October), the last on 28 September, which is some but not most. Version policy is 5 for a first release, and file health is the full 10, with files and card agreeing and no broken-file report waiting as far as the research record shows.

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.

GLM-5.3-Flash 55

File safety is full, with safetensors only, no Python and no remote code needed in Transformers 5.16.1 or later. Provenance is full, files in the zai-org organisation Z.ai links from GitHub with a SHA-256 for every shard. Safety evaluation and agent guidance are 0, as nothing is published on either. Licence risk is the full 15, since MIT adds no policy, threshold or changeable terms. The security programme is 0, with no security.txt on z.ai or docs.z.ai, no SECURITY.md on GitHub and no advisories 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.

GLM-5.3-Flash 70

A Jinja chat template sits in both repositories, for 20. The tool-calling format is 10, because it is defined only by the template and the glm47 parser is named by vLLM and SGLang rather than by Z.ai. The template takes several calls in one turn, for 20. Structured output is 0, with nothing on the card and only the hosted API documented for JSON. Reasoning control is 10, a per-request budget through reasoning_effort, with the reasoning coming back in a think block, though thinking can't be turned off. Context is 10, 1,048,576 tokens native with no rotary scaling.