The Open Weights
LatestModelsLeaderboardsCompanies
Subscribe
The Open Weights

The daily record of open-source AI. New model releases, leaderboards, and what's coming next — written for people who ship.

Refreshed every 12 hours

Discover

  • Latest releases
  • New today
  • Trending models

Browse

  • All models
  • Companies
  • Categories
  • Leaderboards

About

  • About
  • Editorial policy
  • Hardware estimates
  • RSS feed
  • llms.txt
  • Newsletter

© 2026 The Open Weights. An independent publication.

PrivacyTermsSMSAggregated by Claude · curated by humans.
LatestCloudflareclef
CloudflareVision-Language

Cloudflare's Clef brings structured decisions to open models

The new open-weight vision-language family outputs typed, structured results and arrives alongside a reinforcement-learning fine-tuning platform.

Oct 1, 2026
NotableOther
Clef

Cloudflare has introduced Clef, a family of open-weight decision models designed to take image-and-text input and return structured, typed output rather than freeform prose. The release is paired with a new reinforcement-learning fine-tuning platform, according to Cloudflare's announcement.

The pitch is practical: most vision-language models are built to chat, but many real deployments need a model to make a decision and emit a predictable result that downstream software can consume. Clef is framed as a "decision model" aimed at exactly that gap, producing schema-friendly outputs that fit cleanly into automated pipelines.

Why it matters

Typed, structured output is increasingly where the value is for production AI. Teams routinely wrap chat models in parsing and validation layers to coerce reliable results; a model built to return structured decisions natively reduces that brittleness.

  • Open weights published under Cloudflare's clef repository, lowering the barrier to self-hosting and inspection.
  • Multimodal input combining images and text, oriented toward vision-language tasks.
  • An RL fine-tuning platform, giving teams a path to adapt the models to their own decision criteria.

The combination is notable coming from Cloudflare, whose infrastructure footprint positions it to run such models close to where developers already deploy. For teams that need dependable, machine-readable results from multimodal inputs, Clef is worth a look.

Sources

  • Clef: Open-weight decision models, and new RL fine-tuning platform

    Hacker News

    Visit
  • Cloudflare/clef

    Hugging Face

    Visit
OlderFish Audio's Drama 3 brings MoE to open TTSFishaudio · Text → Speech · 2 months agoNewerPerplexity releases a 27B model for multimodal routingPerplexity Ai · Vision-Language · 2 months ago

Get the model

Hugging FaceHacker News

Specs

Context window262K tokens
Size54.7 GB
PrecisionBF16
ArchitectureQwen3_5ForConditionalGeneration
LicenseOTHER
Downloads79K
Likes2.4K

Can you run it?

Runs on a 24 GB GPU at 4-bit.

  • BF16 (as published)

    80 GB GPU (A100 / H100) · 128 GB Mac Studio

    57.8 GB
  • 8-bit

    48 GB GPU (RTX 6000) or 2×24 GB · 64 GB Mac

    32.2 GB
  • 4-bit

    24 GB GPU (RTX 3090 / 4090) · 32 GB Mac

    19.6 GB
Your machine
GGUF builds
  • KV cache sized for 8,192 tokens of context; longer prompts need proportionally more.

Based on Hugging Face weights metadata. Assumes an 8K context and ~1 GB runtime overhead; actual needs vary. How we estimate


Modalities

Any-to-AnyVision-Language

The Weekly Weights

Every open release that mattered, one email a week.

0 comments

No comments yet. Be the first to weigh in.

More from Cloudflare

All Cloudflare releases →
Cloudflare/Text / LLM

Cloudflare releases Clef open-weight decision models

The company pairs its new Clef family with a reinforcement-learning fine-tuning platform aimed at action-oriented tasks.

Oct 1, 2026

More in Vision-Language

All Vision-Language →
SenseNova-Loop-9B
SenseTime/Vision-LanguageRuns on a laptop

SenseTime Releases Looped 9B Multimodal Model

The company's new SenseNova-Loop-9B pairs vision-language capabilities with a recurrent looped-transformer design.

Nov 17, 2026
Hunyuan Hy4-flash
Tencent/Vision-LanguageMulti-GPU server

Tencent releases Hunyuan Hy4-flash multimodal MoE

The lightweight, mixture-of-experts model adds vision and text instruction-following to Tencent's open Hunyuan lineup.

Nov 17, 2026
Mirage
Cactus Compute/Vision-Language

Cactus Compute's Mirage brings vision tasks on-device

The new open vision-language model targets detection, segmentation, and visual Q&A without relying on the cloud.

Nov 5, 2026