github-starred/awesome-open-workhorse-models
2
0
Fork 0
mirror of https://github.com/context-labs/awesome-open-workhorse-models.git synced 2026-10-17 13:03:44 -05:00
Code Issues Packages Projects Releases Wiki Activity

better table

Browse Source
This commit is contained in:
mike
2025-09-15 14:56:17 -07:00
parent 1cb74efa44
commit 6f236fd178
1 changed files with 12 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+12 -12
View File
@@ -17,15 +17,15 @@ Purpose-built, **open-weight** small language (and multimodal) models that enter
## Open Workhorse Models Index
| Model | Task | Companies Using It | Weights/Cards | Key Features |
|-------|------|-------------------|---------------|--------------|
| **Schematron (2025)** | Deterministic HTML ➜ **schema-valid JSON** extraction (long-context) | Financial data aggregators, e-commerce platforms, research firms, content monitoring services | [Schematron-3B (HF)](https://huggingface.co/inference-net/Schematron-3B) · [Ollama card](https://ollama.com/Inference/Schematron) | 3B/8B variants · 128K context · Schema-first extraction · Web data pipelines |
| **ClipTagger-12B (Aug. 2025)** | High-throughput **video/frame tagging** with **structured JSON** | Video streaming platforms, content moderation services, digital asset management companies, advertising networks | [ClipTagger-12B (HF)](https://huggingface.co/inference-net/ClipTagger-12b) | 12B multimodal · FP8 optimized · Enterprise annotation · Cost vs. closed models |
| **ReaderLM-v2 (Jul. 2025)** | **Long-context HTML reading** ➜ Markdown/JSON; layout-aware conversion | Search engines, content aggregators, web scraping services, document processing platforms | [ReaderLM-v2 (HF)](https://huggingface.co/jinaai/ReaderLM-v2) | Crawling pipelines · Document processing · Search stack integration |
| **Foundation-Sec-8B (Cisco) (Apr.–Aug. 2025)** | **Cybersecurity** LLM for SOC workflows: log triage, CTI normalization | Cisco, Splunk, security service providers, enterprise SOC teams, threat intelligence firms | [Foundation-Sec-8B (HF)](https://huggingface.co/fdtn-ai/Foundation-Sec-8B) · [Instruct](https://huggingface.co/fdtn-ai/Foundation-Sec-8B-Instruct) | SOC automation · Incident response · Secure code assist · SIEM/SOAR integration |
| **ShieldGemma 2 (Mar. 2025)** | **Multimodal safety classifier** (image+text) for pre/post-moderation | Social media platforms, content moderation services, online marketplaces, gaming companies | [ShieldGemma 2 (HF)](https://huggingface.co/google/shieldgemma-2-4b-it) · [Transformers docs](https://huggingface.co/docs/transformers/model_doc/shieldgemma2) | NSFW detection · Violence classification · Content moderation · Google/Vertex integration |
| **Llama Guard 4 (Apr. 2025)** | Open-weight **safety/guardrail** classifier (text & multimodal variants) | Enterprise AI platforms, cloud providers, content platforms, educational institutions | [Llama-Guard-4-12B (HF)](https://huggingface.co/meta-llama/Llama-Guard-4-12B) | Safety guardrails · Policy alignment · Enterprise taxonomy · HF provider support |
| **Devstral Small (24B) (May–Jul. 2025)** | **Agentic coding**—multi-file edits, tool use, SWE-Bench Verified SOTA | Enterprise software companies, DevOps platforms, code automation services, AI coding assistants | [Devstral-Small-2505/2507 (HF)](https://huggingface.co/mistralai/Devstral-Small-2507) · GGUF quantizations | SWE-Bench SOTA · Multi-file editing · Tool use · Apache-2.0 license |
| **Qwen3-Coder (Jul.–Sep. 2025)** | **Agentic code generation & long-horizon editing** (extreme context windows) | Enterprise software development teams, code generation platforms, AI coding tools, development agencies | [Qwen3-Coder (HF)](https://huggingface.co/collections/Qwen/qwen3-coder-687fc861e53c939e52d52d10) | 480B MoE variant · Long context · CLI tooling · Vertex AI integration |
| **NV-Embed-v2 (Sep. 2025)** | **Enterprise text embeddings** for RAG—top-tier **MTEB** score | Enterprise search platforms, RAG systems, document management companies, knowledge bases | [NV-Embed-v2 (HF)](https://huggingface.co/nvidia/NV-Embed-v2) | MTEB leader · Long-doc retrieval · Latent attention · NVIDIA NeMo integration |
| **Jina Reranker v2 (Jul. 2025)** | **Cross-encoder reranking** for multilingual search/RAG | Search engines, e-commerce platforms, multilingual content platforms, enterprise search solutions | [jina-reranker-v2-base-multilingual (HF)](https://huggingface.co/jinaai/jina-reranker-v2-base-multilingual) | Multilingual support · AWS Marketplace · Elastic integration · Cross-encoder architecture |
Taking notes on the specialized language models that actually have companies publicly using them. Only including models with concrete adoption evidence, not just platform listings. Focused on real enterprise deployments.
| Model | Companies Actually Using It | Proof of Usage | Blog/Coverage | Key Notes |
|-------|-----------------------------|----------------|---------------|-----------|
| **ClipTagger-12B (Aug. 2025)** | Grass/Wynd Labs | Co-launched the model together, shared training data and ongoing product availability | Launch write-up shows Grass collaboration, model page mentions Wynd partnership | Video tagging specialist that a company co-developed |
| **ReaderLM-v2 (Jul. 2025)** | Jina AI, Google Cloud | Jina AI powers their Reader API with ReaderLM-v2, Google Cloud case study shows Jina Reader at scale on Cloud Run | Jina's model pages list ReaderLM-v2 as the engine, Google Cloud Run case study | HTML-to-Markdown conversion used in production by Jina and scaled on Google Cloud |
| **Foundation-Sec-8B (Cisco) (Apr.–Aug. 2025)** | Cisco, Splunk | Cisco created it for their Foundation AI and internal deployments, Splunk blog about accelerating security ops with it | Cisco Foundation AI blog, Splunk security operations blog | Cybersecurity model with real SOC deployment by Splunk |
| **ShieldGemma 2 (Mar. 2025)** | Google/Vertex AI, Hugging Face | Built into Google's Vertex AI and Gemini as safety model, official implementation in Transformers library | Google/Vertex AI release notes, Transformers model docs | Safety classifier used by Google themselves in their AI stack |
| **Llama Guard 4 (Apr. 2025)** | Databricks, Groq | Listed in Databricks Marketplace for enterprise use, Groq Cloud offers it as production model | Databricks Marketplace listing, Groq Cloud model catalog | Safety guardrails used by major cloud/AI platforms |
| **Qwen3-Coder (Jul.–Sep. 2025)** | Google Cloud, Tinfoil | Available in Vertex AI Model Garden for enterprise deployment, Tinfoil added support to their privacy API | Vertex AI Model Garden docs, Tinfoil API changelog | Coding model with enterprise API integration by Tinfoil |
| **NV-Embed-v2 (Sep. 2025)** | Weaviate, Elastic | Weaviate docs show NVIDIA embedding models available, Elastic Open Inference API adds NVIDIA text-embedding models | Weaviate documentation, Elastic Open Inference API announcement | RAG embeddings used by vector database companies |
| **Jina Reranker v2 (Jul. 2025)** | Elastic, AWS | Elastic announces Jina reranker support in Open Inference API, AWS Marketplace lists Jina Reranker for enterprise deployment | Elastic Open Inference API blog, AWS Marketplace listing | Search reranking used by enterprise search platforms |