OpenModels
ModelsProvidersSkillsMCPCompareInsights
DocsGitHub
OpenModels

Open-source registry for LLM models, inference providers, and AI agent skills. Community-maintained.

[email protected]

Registry

  • Models
  • Providers
  • Skills
  • MCP
  • Collections
  • Compare

Resources

  • Documentation
  • API Reference
  • CLI
  • Contribute
  • Contributing

Community

  • GitHub
  • Issues
  • Discussions

© 2026 OpenModels. Open-source, community-maintained.

Privacy PolicyAll systems operational
OpenModelsOpenModels

OpenModels comparison

Choose the right AI model for your workload

Filter the full registry, compare provider-specific prices and capacity, then estimate cost with your own assumptions. Missing performance data stays missing.

Filters and result options

Filters 1

Global model explorer

50 matching models

Contexttokens

Canonical maximum token capacity. Capacity is not quality.

Model shortlist

Flagship and recently added models

Editorial flagship labels and registry dates affect discovery order, never model scores.

Models with registry facts and provider-attributed pricing
#ModelDeveloper / countrySelected metricProvider price · in / outContextRegistry dateCompare
1
DeepSeek R1flagship
open weights · unknown
Deepseekcn131K$0.40 / $1.60Deep Infra131K1/20/2025Official registry
2
GLM-5.2recent
open weights · unknown
Zhipu AIcn1M$0.60 / $2.00OpenRouter1M6/13/2026Official registry
3
Sarvam-Mrecent
open weights · unknown
Sarvamin131K$0.25 / $0.75OpenRouter131K6/13/2026Official registry
4
Sarvam-1recent
open weights · unknown
Sarvamin8K$0.10 / $0.20Sarvam AI8K6/13/2026Official registry
5
Sarvam-105Brecent
open weights · unknown
Sarvamin128K$1.00 / $3.00Sarvam AI128K6/13/2026Official registry
6
Sarvam-30Brecent
open weights · unknown
Sarvamin128K$0.60 / $1.80Sarvam AI128K6/13/2026Official registry
7
DiffusionGemmarecent
open weights · unknown
Diffusiongemmaus262K$0.15 / $0.15Hugging Face Inference262K6/11/2026Official registry
8
Gemma 4 12Brecent
open weights · unknown
Googleus262K$0.10 / $0.10Hugging Face Inference262K6/4/2026Official registry
9
Falcon-H1recent
open weights · unknown
Falconae131K$0.50 / $1.50OpenRouter131K5/25/2026Official registry
10
Falcon 3 10Brecent
open weights · unknown
Falconae33K$0.10 / $0.20Lambda33K5/25/2026Official registry
11
Ring-2.6-1Trecent
open weights · unknown
Ringcn131KNot priced131K5/25/2026Official registry
12
Snowflake Arcticrecent
open weights · unknown
Snowflakeus4KNot priced4K5/25/2026Official registry
13
K2 Thinkrecent
open weights · unknown
K2ae131K$0.60 / $0.60Cerebras131K5/23/2026Official registry
14
MiniCPM-V 4.6recent
open weights · unknown
Minicpmcn256KNot priced256K5/8/2026Official registry
15
Granite 4.1 8Brecent
open weights · unknown
IBMus131K$0.05 / $0.10OpenRouter131K5/1/2026Official registry
16
Granite 4.1 30Brecent
open weights · unknown
IBMus524K$0.60 / $1.20IBM watsonx.ai524K5/1/2026Official registry
17
DeepSeek V4 Flashrecent
open weights · unknown
Deepseekcn1M$0.07 / $0.14Deep Infra1M4/24/2026Official registry
18
DeepSeek V4 Prorecent
open weights · unknown
Deepseekcn1M$0.55 / $2.19DeepSeek1M4/24/2026Official registry
19
MiMo-V2.5-Prorecent
open weights · unknown
Mimocn1M$1.00 / $3.00OpenRouter1M4/22/2026Official registry
20
Qwen 3.6 27Brecent
open weights · unknown
Qwen3cn131K$0.20 / $0.60Alibaba Model Studio131K4/22/2026Official registry
Page 1 of 3

Selected models

Side-by-side comparison

Compare capacity, deployment choice and published cost without synthetic quality scores.

Choose 2–4 models
Choose at least two models from the chart or shortlist.

Cost per task

Estimate your workload, not an abstract blend

Uses the named provider price and your token assumptions. Tool, image, audio and unentered reasoning-token charges are excluded.

Short multi-turn chat; no cache assumed.

Select models with current provider pricing to estimate cost.

Data trust

What this version can—and cannot—claim

Available

Registry facts

Canonical metadata, capabilities, context, access classification, provider deployments and current published prices.

Not evaluated

Quality & benchmarks

No composite score is shown. Only versioned, sourced benchmark records will be admitted.

Not measured

Speed & latency

Legacy probes are excluded until provider, region, prompt, sample size and methodology are comparable.

Read methodology and data policy →