GLM-5.3
Z.ai China
Best overall GLM-5.3 License
Z.ai's strongest open-weight model, post-trained for complex coding and long-horizon agent work.
- Context
- 1M tokens
- Params
- 753B
- Input
- text
The best open source AI models, ranked and compared
The top open-weight models for AI agents, with specs, licenses, and where each one is from.
Updated · 15 models 10 from China 5 from the US
The best open source LLMs for agents
Ranked for agent work: tool use, context, license, and cost at volume.
Kimi K3
Moonshot AI China
Most capable Kimi K3 License
The first open 3T-class model, natively multimodal with a 1M-token context.
- Context
- 1M tokens
- Params
- 2.8T
- Input
- text, image, video
DeepSeek V4 Pro
DeepSeek China
Best for reasoning MIT
The official DeepSeek V4 Pro release, with stronger agentic skills and an MIT license.
- Context
- 1M tokens
- Params
- 1.6T
- Input
- text
Muse Spark Open
Meta US
Best US model Muse Model License
Meta's open-weight Muse Spark, built for coding and agent work with a 1M-token context.
- Context
- 1M tokens
- Params
- 816B · 42B active
- Input
- text, image, audio, video
GLM-5.3 Flash
Z.ai China
Best value MIT
A natively multimodal GLM that Z.ai says beats GLM-5.2 at one-tenth the price.
- Context
- 1M tokens
- Params
- 320B · 18B active
- Input
- text, image
DeepSeek V4.1 Flash
DeepSeek China
Fastest MIT
A multimodal MoE that activates only 8B parameters per token and shrinks the KV cache about 4x.
- Context
- 1M tokens
- Params
- 552B · 8B active
- Input
- text, image
Kimi K2.7 Code
Moonshot AI China
Best for coding Modified MIT
A coding-focused Kimi tuned for long-horizon software engineering.
- Context
- 256K tokens
- Params
- 1T · 32B active
- Input
- text, image
Nemotron 3.5 Super VL
NVIDIA US
Best for documents and screens OpenMDW-1.1
NVIDIA's vision-language agent model, with 12B active parameters and a 1M-token context.
- Context
- 1M tokens
- Params
- 120B · 12B active
- Input
- text, image, video
Qwen3.8 2.4T-A95B
Qwen (Alibaba) China
Best for multilingual Qwen3.8-Max License
The first Qwen-Max-class model released with open weights.
- Context
- 256K tokens
- Params
- 2.4T · 95B active
- Input
- text
MiniMax M3
MiniMax China
Best multimodal MiniMax Community License
A natively multimodal model with sparse attention built for million-token contexts.
- Context
- 1M tokens
- Params
- 428B · 23B active
- Input
- text, image, video
Qwen3.8 27B
Qwen (Alibaba) China
Best to self-host Apache 2.0
A compact dense vision-language model under Apache 2.0.
- Context
- 256K tokens
- Params
- 27B
- Input
- text, image, video
Muse Glimmer
Meta US
Best for local agents Apache 2.0
A 30B agent model distilled from Muse Spark that runs on one 24GB to 32GB GPU or a Mac.
- Context
- 128K tokens
- Params
- 30B
- Input
- text, image
Nemotron 3.5 Bolt
NVIDIA US
Best on a 24GB GPU OpenMDW-1.1
A 74B model with 9B active that NVIDIA also ships in a 2-bit build sized for 24GB.
- Context
- 256K tokens
- Params
- 74B · 9B active
- Input
- text, image, video
Qwen3.8 Flash Next
Qwen (Alibaba) China
Most efficient long context Qwen Community License
An experimental preview of the architecture that will underpin Qwen4.
- Context
- 256K tokens
- Params
- 180B
- Input
- text, image
gpt-oss-120b
OpenAI US
Best on a single GPU Apache 2.0
OpenAI's open-weight reasoning model, small enough to run on a single 80GB GPU.
- Context
- 128K tokens
- Params
- 117B · 5.1B active
- Input
- text
Benchmark scores are reported by each model provider on its Hugging Face model card. They are not independently verified by Runtime.
Open source models compared
Filter by license, input, or origin. Sort by context or release date.
Filter:Origin:15 of 15 models
| Origin | Parameters | License | Input | Good for | ||||
|---|---|---|---|---|---|---|---|---|
| 1 | GLM-5.3Best overall | China | 753B | 1M tokens | GLM-5.3 License | text | Aug 2026 | Payment reconciliationChargebacks and disputesMerchant underwriting |
| 2 | Kimi K3Most capable | China | 2.8T | 1M tokens | Kimi K3 License | text, image, video | Jun 2026 | Partner-bank RFIsChargebacks and disputesMerchant underwriting |
| 3 | DeepSeek V4 ProBest for reasoning | China | 1.6T | 1M tokens | MIT | text | Aug 2026 | Payment reconciliationMerchant underwritingAML and compliance investigations |
| 4 | Muse Spark OpenBest US model | US | 816B / 42B | 1M tokens | Muse Model License | text, image, audio, video | Oct 2026 | |
| 5 | GLM-5.3 FlashBest value | China | 320B / 18B | 1M tokens | MIT | text, image | Aug 2026 | Partner-bank RFIsTransaction monitoringKYC and KYB checks |
| 6 | DeepSeek V4.1 FlashFastest | China | 552B / 8B | 1M tokens | MIT | text, image | Sep 2026 | Transaction monitoringKYC and KYB checksPayments customer support |
| 7 | Kimi K2.7 CodeBest for coding | China | 1T / 32B | 256K tokens | Modified MIT | text, image | Jun 2026 | |
| 8 | Nemotron 3.5 Super VLBest for documents and screens | US | 120B / 12B | 1M tokens | OpenMDW-1.1 | text, image, video | Oct 2026 | |
| 9 | Qwen3.8 2.4T-A95BBest for multilingual | China | 2.4T / 95B | 256K tokens | Qwen3.8-Max License | text | Aug 2026 | AML and compliance investigationsPayments customer support |
| 10 | MiniMax M3Best multimodal | China | 428B / 23B | 1M tokens | MiniMax Community License | text, image, video | Jun 2026 | Partner-bank RFIsChargebacks and disputes |
| 11 | Qwen3.8 27BBest to self-host | China | 27B | 256K tokens | Apache 2.0 | text, image, video | Aug 2026 | Payment reconciliationKYC and KYB checks |
| 12 | Muse GlimmerBest for local agents | US | 30B | 128K tokens | Apache 2.0 | text, image | Aug 2026 | |
| 13 | Nemotron 3.5 BoltBest on a 24GB GPU | US | 74B / 9B | 256K tokens | OpenMDW-1.1 | text, image, video | Oct 2026 | |
| 14 | Qwen3.8 Flash NextMost efficient long context | China | 180B | 256K tokens | Qwen Community License | text, image | Aug 2026 | |
| 15 | gpt-oss-120bBest on a single GPU | US | 117B / 5.1B | 128K tokens | Apache 2.0 | text | Aug 2025 | Transaction monitoring |
Parameters show total / active per token for mixture-of-experts models.
Open source models by country
Weights run wherever you host them. Origin still comes up in vendor and model risk reviews, so here is where each provider is based.
China10 models
Best open source models by use case
Our picks for the work payment teams hand to agents.
Payment ops Risk and fraud Underwriting Compliance Support
Payment opsBest open source models for payment reconciliationTop pick: DeepSeek V4 ProSee the picksPayment opsBest open source models for partner-bank RFIsTop pick: Kimi K3See the picksRisk and fraudBest open source models for transaction monitoringTop pick: DeepSeek V4.1 FlashSee the picksRisk and fraudBest open source models for chargebacks and disputesTop pick: MiniMax M3See the picksUnderwritingBest open source models for merchant underwritingTop pick: GLM-5.3See the picksUnderwritingBest open source models for KYC and KYB checksTop pick: Qwen3.8 27BSee the picksComplianceBest open source models for AML and compliance investigationsTop pick: DeepSeek V4 ProSee the picksSupportBest open source models for payments customer supportTop pick: GLM-5.3 FlashSee the picks
Top pick: DeepSeek V4 ProSee the picksPayment opsBest open source models for partner-bank RFIsTop pick: Kimi K3See the picksRisk and fraudBest open source models for transaction monitoringTop pick: DeepSeek V4.1 FlashSee the picksRisk and fraudBest open source models for chargebacks and disputesTop pick: MiniMax M3See the picksUnderwritingBest open source models for merchant underwritingTop pick: GLM-5.3See the picksUnderwritingBest open source models for KYC and KYB checksTop pick: Qwen3.8 27BSee the picksComplianceBest open source models for AML and compliance investigationsTop pick: DeepSeek V4 ProSee the picksSupportBest open source models for payments customer supportTop pick: GLM-5.3 FlashSee the picksOpen source vs frontier models
Frontier models lead on the hardest tasks. Open models win on cost and control.
| Dimension | Open source models | Frontier models |
|---|---|---|
| Cost | A fraction of frontier pricing | Premium on top tiers |
| Data control | Runs in your cloud | Provider's data policy |
| Customization | Fine-tune the weights | Prompting only |
| Peak capability | Close behind | Still leads on the hardest tasks |
| Hosting | Self-host, Bedrock, or Vertex AI | Provider API only |
Run open models inside your own cloud
Pick a model per agent or per skill. Evals tell you when a cheaper one is good enough.
Your cloud
Bedrock, Vertex AI, or your own vLLM or SGLang endpoint. Card data stays put.
Any harness
OpenCode, Cline, or Claude Code with an Anthropic-compatible API.
Routed by evals
Move routine work to the smallest model that still passes.
Understanding open source AI models
What are open source AI models?
Language models whose weights you can download, run, and usually modify. Host them yourself, on Bedrock or Vertex AI, or inside Runtime.
Open weight vs open source
Most are open weight: the weights are public, the training data is not. MIT, Apache 2.0, and OpenMDW are permissive. Custom licenses can add conditions.
Why mixture-of-experts is cheap
Each token uses a small slice of the network, so an 800B model may only run 40B parameters per token.
How to choose
Match context to your inputs, check the license, then test a few models on your real cases.
Open source AI models FAQ
What is the best open source AI model right now?
+
As of October 16, 2026, GLM-5.3 from Z.ai. Z.ai's strongest open-weight model, post-trained for complex coding and long-horizon agent work.
Which open source model is best for coding?
+
Kimi K2.7 Code from Moonshot AI. A coding-focused Kimi tuned for long-horizon software engineering.
What is the best open source model made in the US?
+
Muse Spark Open from Meta ranks highest. The other US models in our list are Nemotron 3.5 Super VL (NVIDIA), Muse Glimmer (Meta), Nemotron 3.5 Bolt (NVIDIA), gpt-oss-120b (OpenAI).
Which open source models can I use commercially?
+
MIT, Apache 2.0, and OpenMDW allow commercial use and fine-tuning: DeepSeek V4 Pro, GLM-5.3 Flash, DeepSeek V4.1 Flash, Nemotron 3.5 Super VL, Qwen3.8 27B, Muse Glimmer, Nemotron 3.5 Bolt, gpt-oss-120b. The rest use custom licenses, so read the terms first.
Are open source models as good as frontier models?
+
Frontier models still lead on the hardest tasks. Open models are close on most agent benchmarks and win on cost, privacy, and control.
What is the difference between open source and open weight?
+
Open weight means the trained weights are public. Fully open source also publishes the training data and code. Most "open source" models are open weight, so check the license.
Which open source model is best to self-host?
+
Qwen3.8 27B. A compact dense vision-language model under Apache 2.0. For one 80GB GPU, try gpt-oss-120b.
Can I run these open source models in Runtime?
+
Yes. Runtime is model-neutral. Serve a model from Bedrock, Vertex AI, or your own endpoint, and use evals to move routine work to cheaper models.
Model profiles
GLM-5.3 Z.ai · Best overallKimi K3 Moonshot AI · Most capableDeepSeek V4 Pro DeepSeek · Best for reasoningMuse Spark Open Meta · Best US modelGLM-5.3 Flash Z.ai · Best valueDeepSeek V4.1 Flash DeepSeek · FastestKimi K2.7 Code Moonshot AI · Best for codingNemotron 3.5 Super VL NVIDIA · Best for documents and screensQwen3.8 2.4T-A95B Qwen (Alibaba) · Best for multilingualMiniMax M3 MiniMax · Best multimodalQwen3.8 27B Qwen (Alibaba) · Best to self-hostMuse Glimmer Meta · Best for local agentsNemotron 3.5 Bolt NVIDIA · Best on a 24GB GPUQwen3.8 Flash Next Qwen (Alibaba) · Most efficient long contextgpt-oss-120b OpenAI · Best on a single GPURun open source models in Runtime
Pick a model and put an agent to work in your own cloud.