Mistral Medium

weights

open(safetensors on HF mistralai/Mistral-Medium-3.5-128B, ungated)

data

closed

code

partial(inference and serving only

license

Mistral-Modified-MIT(non-OSI

Mistral Medium 3.5's weights are downloadable without a gate, but its Modified MIT license withholds every right from companies with more than $20 million in monthly revenue, and Mistral publishes neither the training data nor the training code. That makes it an open-weights release under a bounded license.

• https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B recorded 2026-10-12

  The Hugging Face API record reports gated as false, carries the tag "license:other", and lists weight shards from "consolidated-00001-of-00003.safetensors" to "consolidated-00003-of-00003.safetensors". It declares no dataset tag.

• https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/raw/main/LICENSE recorded 2026-10-12

  LICENSE file, "Modified MIT License": "You are not authorized to exercise any rights under this license if the global consolidated monthly revenue of your company (or that of your employer) exceeds $20 million (or its equivalent in another currency) for the preceding month." A revenue ceiling, so not OSI.

• https://huggingface.co/mistralai/Mistral-Medium-3.5-128B/raw/main/README.md recorded 2026-10-12

  Raw model card front matter declares "license: other", and the card says "We release this model under a Modified MIT License: Open-source license for both commercial and non-commercial use with exceptions for companies with large revenue". Searched the body for dataset, training data and corpus: none. The code it references is for serving (vLLM, mistral_common), not training.

Adoption is measured by Hugging Face downloads of the one checkpoint Mistral publishes. Mistral also serves the model through its API and Le Chat, which publish no usage figure, so the count covers the open weights alone.

• https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B recorded 2026-10-12

  72,266 downloads in the trailing 30 days for mistralai/Mistral-Medium-3.5-128B

Mistral Medium 3.5 is among the strongest open-weight models of its size on the Artificial Analysis index, but it sits far below the frontier, between GPT-4o and o3. That places it level with OpenAI's o-series, a step below Gemini 3.1 Pro.

• https://artificialanalysis.ai/leaderboards/models recorded 2026-10-12

  The leaderboard's embedded model records give intelligenceIndex 14.19 for Mistral Medium 3.5, 29.72 for Gemini 3.1 Pro Preview, 9.99 for Llama 4 Maverick, 9.27 for Mistral Large 3 and 8.44 for GPT-4o (Nov '24). The digest drifts between fetches because the page carries live pricing and throughput figures.

• https://artificialanalysis.ai/models/mistral-medium-3-5 recorded 2026-10-12

  "Mistral Medium 3.5 scores 14 on the Artificial Analysis Intelligence Index, placing it well above average among comparable models (median: 8)"; Intelligence rank "# 6 / 65"; the page tags it "Open weights model" and "Released April 2026".

• https://docs.mistral.ai/models/mistral-medium-3-5-26-04 recorded 2026-10-12

  Mistral's model page: "April 28, 2026", "Modified MIT", "v 26.04", "Mistral Medium 3.5. Our frontier-class multimodal model optimized for agentic and coding use cases. Released as open weights under a Modified MIT license", context 256k, with its weights linked at huggingface.co/mistralai/Mistral-Medium-3.5-128B.