Veyrul

sourceactiveImage generation

HiDream-ai/HiDream-I1

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 1 · updated 2026-10-18 · src_01M569CCGZ5AZRQ9ZDXGG2H3EE

Evidence: review revision unknown · inspection metadata_checked · execution not_run · review unreviewed · origin research_agent

Source: https://github.com/hidream-ai/hidream-i1

Structured fields
{
  "archived": false,
  "checked_at": "2026-10-17T00:00:00.000Z",
  "usefulness": "Official repo for HiDream-I1, a 17B-parameter open text-to-image model with Full (50 steps), Dev (28, distilled) and Fast (16, distilled) variants, released under MIT per the README. Notable for MIT licensing of both code and models; its default text-encoder path depends on a separately licensed Llama-3.1-8B-Instruct.",
  "source_type": "repository",
  "availability": "available",
  "dependencies": [
    "Llama-3.1-8B-Instruct (separate license)",
    "flash-attn (recommended)",
    "CUDA 12.4 (suggested)"
  ],
  "source_owner": "HiDream.ai",
  "next_check_at": "2026-10-31T00:00:00.000Z",
  "known_limitations": [
    "No VRAM requirements stated.",
    "Llama text-encoder license is a separate dependency requirement.",
    "HiDream-E1-1 editing model not inspected."
  ],
  "upstream_updated_at": "2025-07-16T14:38:13.000Z"
}

Evidence claims

  • [sourced_fact] README: code and HiDream-I1 models are both under the MIT License; Full 50 steps, Dev 28 steps, Fast 16 steps; Dev/Fast use guidance 0.0, Full 5.0 in the Diffusers example. source
  • [sourced_fact] Automatic text-encoder download requires accepting the Meta Llama-3.1-8B-Instruct license on Hugging Face; Flash Attention and CUDA 12.4 suggested. source
  • [sourced_fact] Latest dated entry on the page is 16 July 2025, announcing open-sourcing of the editing model HiDream-E1-1. source
  • [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.

Rights

componentlicensehostingredistributionmodificationtrainingEvidence and notes
codeMITpermittedpermittedpermittedunknown

README states MIT; LICENSE file not separately fetched.

model_weightsMITpermittedpermittedpermittedunknown

README states the models are MIT; Hugging Face model card not fetched. Llama-3.1-8B-Instruct text encoder has its own Meta license (not read).

Machine-readable: JSON · Markdown · revisions · changes