What is the best open source models product?

Anthropic admits to have made hosted models more stupid, proving the importance of open weight, local models is the top-rated open source models product with a score of 49/100 on Unfragile.

How many open source models products are there?

Unfragile indexes 15 open source models products, ranked by real usage data.

Best Open Source Models (2026) | Unfragile

Hub

Browse all artifacts

Agents Models MCP Servers Frameworks

Home/Categories/Open Source Models

Open Source Models

15 tools · Browse 15 open source models AI artifacts on Unfragile.

Avg score: 44/100

Anthropic admits to have made hosted models more stupid, proving the importance of open weight, local models

49/100Paid

Anthropic admits to have made hosted models more stupid, proving the importance of open weight, local models

Model

Gemma 4 just casually destroyed every model on our leaderboard except Opus 4.6 and GPT-5.2. 31B params, $0.20/run

Search

Search AI Artifacts
For Developers
For Idea Builders
Categories
Trends
Compare
Stacks
Use Cases

Hub

Browse All
Capabilities
Agents
Models
MCP Servers
Repositories

For Builders

Submit an Artifact
Studio Dashboard
Demand Gaps
What's Missing
Capability Schema

Discover

Capabilities Trends Compare Stacks Use Cases Categories Demand Gaps

Studio

API

For Builders

49/100

Paid

Gemma 4 just casually destroyed every model on our leaderboard except Opus 4.6 and GPT-5.2. 31B params, $0.20/run

Model

Qwen 3.6 27B is out

47/100Paid

Qwen 3.6 27B is out

Model

Claude Code removed from Claude Pro plan - better time than ever to switch to Local Models.

47/100Paid

Claude Code removed from Claude Pro plan - better time than ever to switch to Local Models.

Model

Qwen3.6-35B-A3B released!

47/100Paid

Qwen3.6-35B-A3B released!

Model

Deepseek v4 people

46/100Paid

Deepseek v4 people

Model

Forgive my ignorance but how is a 27B model better than 397B?

45/100Paid

Forgive my ignorance but how is a 27B model better than 397B?

Model

You can now fine-tune Gemma 4 locally 8GB VRAM + Bug Fixes

45/100Paid

You can now fine-tune Gemma 4 locally 8GB VRAM + Bug Fixes

Model

Deepseek V4 Flash and Non-Flash Out on HuggingFace

41/100Paid

Deepseek V4 Flash and Non-Flash Out on HuggingFace

Model

Qwen3.6-27B released!

41/100Paid

Qwen3.6-27B released!

Model

Kimi K2.6 Released (huggingface)

41/100Paid

Kimi K2.6 Released (huggingface)

Model

Minimax M2.7 Released

41/100Paid

Minimax M2.7 Released

Model

Gemma 4 31B beats several frontier models on the FoodTruck Bench

40/100Paid

Gemma 4 31B beats several frontier models on the FoodTruck Bench

Model

Qwen3.6. This is it.

39/100Paid

Qwen3.6. This is it.

1-bit Bonsai 1.7B (290MB in size) running locally in your browser on WebGPU

38/100Paid

1-bit Bonsai 1.7B (290MB in size) running locally in your browser on WebGPU

Web App

← All categories Search tools

API Docs

Company

About
Philosophy

nfragile