wehub-resource-sync
94057c3d3e
PR Test (NPU) / check-changes (push) Has been cancelled
PR Test (NPU) / pr-gate (push) Has been cancelled
PR Test (NPU) / set-image-config (push) Has been cancelled
PR Test (NPU) / stage-b-test-1-npu-a2 (0) (push) Has been cancelled
PR Test (NPU) / stage-b-test-1-npu-a2 (1) (push) Has been cancelled
PR Test (NPU) / stage-b-test-2-npu-a2 (0) (push) Has been cancelled
PR Test (NPU) / stage-b-test-2-npu-a2 (1) (push) Has been cancelled
PR Test (NPU) / stage-b-test-4-npu-a3 (push) Has been cancelled
PR Test (NPU) / stage-b-test-16-npu-a3 (push) Has been cancelled
PR Test (NPU) / multimodal-gen-test-1-npu-a3 (push) Has been cancelled
PR Test (NPU) / multimodal-gen-test-2-npu-a3 (push) Has been cancelled
PR Test (Arm64) / pr-gate (push) Has been cancelled
PR Test (Arm64) / check-changes (push) Has been cancelled
PR Test (Arm64) / build-test (push) Has been cancelled
PR Test (sgl-router) / gate (push) Has been cancelled
PR Test (sgl-router) / tier-1 — lint (push) Has been cancelled
PR Test (sgl-router) / tier-2 — build + test (push) Has been cancelled
PR Test (sgl-router) / tier-3 — docker (placeholder) (push) Has been cancelled
PR Test (sgl-router) / tier-3 — k8s integration (push) Has been cancelled
PR Test (sgl-router) / tier-3 — e2e (push) Has been cancelled
PR Test (sgl-router) / finish (push) Has been cancelled
PR Test (NPU) / single-node-poc (map[name:qwen3_6_27b_w8a8_1p_in64k_out1k_50ms runner:linux-aarch64-a3-2 test_case:test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_1p_in64k_out1k_50ms.py test_type:perf]) (push) Has been cancelled
PR Test (NPU) / pr-test-npu-finish (push) Has been cancelled
PR Test (Xeon) / pr-gate (push) Has been cancelled
PR Test (Xeon) / check-changes (push) Has been cancelled
PR Test (Xeon) / build-test (, xeon-gnr, base-b-test-cpu) (push) Has been cancelled
PR Test (XPU) / check-changes (push) Has been cancelled
PR Test (XPU) / pr-gate (push) Has been cancelled
PR Test (XPU) / stage-a-test-1-gpu-xpu (push) Has been cancelled
PR Test (XPU) / wait-for-stage-a (push) Has been cancelled
PR Test (XPU) / stage-b-test-1-gpu-xpu (push) Has been cancelled
PR Test (XPU) / finish (push) Has been cancelled
CI Model Inventory / build-inventory (push) Has been cancelled
Lint / lint (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark Compilation Check (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark - Manual Policy (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark - Request Processing (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark Summary (push) Has been cancelled
PR Test (SMG) / build-wheel (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on windows (x86_64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on macos (x86_64 - auto) (push) Has been cancelled
PR Test (SMG) / python-unit-tests (push) Has been cancelled
PR Test (SMG) / unit-tests (push) Has been cancelled
PR Test (SMG) / benchmarks (push) Has been cancelled
PR Test (SMG) / chat-completions (push) Has been cancelled
PR Test (SMG) / chat-completions-4gpu (push) Has been cancelled
PR Test (SMG) / e2e (push) Has been cancelled
PR Test (SMG) / docker-build-test (push) Has been cancelled
PR Test (SMG) / k8s-integration (push) Has been cancelled
PR Test (SMG) / finish (push) Has been cancelled
PR Test (SMG) / summarize-benchmarks (push) Has been cancelled
Release SGLang Model Gateway Docker Image / publish (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on macos (aarch64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (aarch64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (x86_64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (aarch64 - musllinux_1_1) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (x86_64 - musllinux_1_1) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / Build SDist (push) Has been cancelled
Release SGLang Model Gateway to PyPI / Upload to PyPI (push) Has been cancelled
Release SGLang Kernels / build-cu129-matrix (aarch64, 12.9, 3.10, arm-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / build-cu129-matrix (x86_64, 12.9, 3.10, x64-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / release-cu129 (push) Has been cancelled
Release SGLang Kernels / build-cu130-matrix (aarch64, 13.0, 3.10, arm-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / build-cu130-matrix (x86_64, 13.0, 3.10, x64-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / release-cu130 (push) Has been cancelled
Release SGLang Kernels / build-rocm-matrix (3.10, 700) (push) Has been cancelled
Release SGLang Kernels / build-rocm-matrix (3.10, 720) (push) Has been cancelled
Release SGLang Kernels / release-rocm700 (push) Has been cancelled
Release SGLang Kernels / release-rocm720 (push) Has been cancelled
Release SGLang Kernels / build-musa43 (43, 3.10) (push) Has been cancelled
Release SGLang Kernels / release-musa43 (push) Has been cancelled
293 lines
28 KiB
Plaintext
293 lines
28 KiB
Plaintext
---
|
||
title: Large Language Models
|
||
---
|
||
These models accept text input and produce text output (e.g., chat completions). They are primarily large language models (LLMs), some with mixture-of-experts (MoE) architectures for scaling.
|
||
|
||
## Example launch Command
|
||
|
||
```shell Command
|
||
python3 -m sglang.launch_server \
|
||
--model-path meta-llama/Llama-3.2-1B-Instruct \ # example HF/local path
|
||
--host 0.0.0.0 \
|
||
--port 30000 \
|
||
```
|
||
|
||
## Supported models
|
||
|
||
Below the supported models are summarized in a table.
|
||
|
||
If you are unsure if a specific architecture is implemented, you can search for it via GitHub. For example, to search for `Qwen3ForCausalLM`, use the expression:
|
||
|
||
```text Output
|
||
repo:sgl-project/sglang path:/^python\/sglang\/srt\/models\// Qwen3ForCausalLM
|
||
```
|
||
|
||
in the GitHub search bar.
|
||
|
||
<table style={{width: "100%", borderCollapse: "collapse", tableLayout: "fixed"}}>
|
||
<colgroup>
|
||
<col style={{width: "34%"}} />
|
||
<col style={{width: "33%"}} />
|
||
<col style={{width: "33%"}} />
|
||
</colgroup>
|
||
<thead>
|
||
<tr style={{borderBottom: "2px solid #d55816"}}>
|
||
<th style={{textAlign: "left", padding: "10px 12px", fontWeight: 700, whiteSpace: "nowrap", backgroundColor: "rgba(255,255,255,0.02)"}}>Model Family (Variants)</th>
|
||
<th style={{textAlign: "left", padding: "10px 12px", fontWeight: 700, whiteSpace: "nowrap", backgroundColor: "rgba(255,255,255,0.05)"}}>Example HuggingFace Identifier</th>
|
||
<th style={{textAlign: "left", padding: "10px 12px", fontWeight: 700, whiteSpace: "nowrap", backgroundColor: "rgba(255,255,255,0.02)"}}>Description</th>
|
||
</tr>
|
||
</thead>
|
||
<tbody>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**DeepSeek** (v1, v2, v3/R1)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`deepseek-ai/DeepSeek-R1`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Series of advanced reasoning-optimized models (including a 671B MoE) trained with reinforcement learning; top performance on complex reasoning, math, and code tasks. <a href="/cookbook/autoregressive/DeepSeek/DeepSeek-V3">SGLang provides Deepseek v3/R1 model-specific optimizations</a> and <a href="../advanced_features/separate_reasoning">Reasoning Parser</a></td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Kimi K2** (Thinking, Instruct)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`moonshotai/Kimi-K2-Instruct`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Moonshot AI's 1 trillion parameter MoE model (32B active) with 128K–256K context; state-of-the-art agentic intelligence with stable long-horizon agency across 200–300 sequential tool calls. Features MLA attention and native INT4 quantization. <a href="../advanced_features/separate_reasoning">See Reasoning Parser docs</a></td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Kimi Linear** (48B-A3B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`moonshotai/Kimi-Linear-48B-A3B-Instruct`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Moonshot AI's hybrid linear attention model (48B total, 3B active) with 1M token context; features Kimi Delta Attention (KDA) for up to 6× faster decoding and 75% KV cache reduction vs full attention.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**GPT-OSS**</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>openai/gpt-oss-20b</code>, <code>openai/gpt-oss-120b</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>OpenAI’s latest GPT-OSS series for complex reasoning, agentic tasks, and versatile developer use cases.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>Qwen</strong> (3.5, 3, 3MoE, 3Next, 2.5, 2 series)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>Qwen/Qwen3.5-397B-A17B</code>, <code>Qwen/Qwen3-0.6B</code>, <code>Qwen/Qwen3-30B-A3B</code>, <code>Qwen/Qwen3-Next-80B-A3B-Instruct</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Alibaba’s latest Qwen3 series for complex reasoning, language understanding, and generation tasks; Support for MoE variants along with previous generation 2.5, 2, etc. <a href="../advanced_features/separate_reasoning">SGLang provides Qwen3 specific reasoning parser</a></td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Llama** (2, 3.x, 4 series)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`meta-llama/Llama-4-Scout-17B-16E-Instruct`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Meta's open LLM series, spanning 7B to 400B parameters (Llama 2, 3, and new Llama 4) with well-recognized performance. <a href="/cookbook/autoregressive/Llama/Llama4">SGLang provides Llama-4 model-specific optimizations</a></td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Mistral** (Mixtral, NeMo, Small3)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`mistralai/Mistral-7B-Instruct-v0.2`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Open 7B LLM by Mistral AI with strong performance; extended into MoE (“Mixtral”) and NeMo Megatron variants for larger scale.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Gemma** (v1, v2, v3)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`google/gemma-3-1b-it`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Google’s family of efficient multilingual models (1B–27B); Gemma 3 offers a 128K context window, and its larger (4B+) variants support vision input.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Phi** (Phi-1.5, Phi-2, Phi-3, Phi-4, Phi-MoE series)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>microsoft/Phi-4-multimodal-instruct</code>, <code>microsoft/Phi-3.5-MoE-instruct</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Microsoft’s Phi family of small models (1.3B–5.6B); Phi-4-multimodal (5.6B) processes text, images, and speech, Phi-4-mini is a high-accuracy text model and Phi-3.5-MoE is a mixture-of-experts model.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**MiniCPM** (v3, 4B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`openbmb/MiniCPM3-4B`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>OpenBMB’s series of compact LLMs for edge devices; MiniCPM 3 (4B) achieves GPT-3.5-level results in text tasks.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**OLMo** (2, 3)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>allenai/OLMo-3-1125-32B</code>, <code>allenai/OLMo-3-32B-Think</code>, <code>allenai/OLMo-2-1124-7B-Instruct</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Allen AI’s series of Open Language Models designed to enable the science of language models.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**OLMoE** (Open MoE)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`allenai/OLMoE-1B-7B-0924`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Allen AI’s open Mixture-of-Experts model (7B total, 1B active parameters) delivering state-of-the-art results with sparse expert activation.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>MiniMax-M2</strong> (M2, M2.1, M2.5)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>MiniMaxAI/MiniMax-M2.5</code>, <code>MiniMaxAI/MiniMax-M2.1</code>, <code>MiniMaxAI/MiniMax-M2</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>MiniMax's SOTA LLM for coding & agentic workflows.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**StableLM** (3B, 7B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`stabilityai/stablelm-tuned-alpha-7b`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>StabilityAI’s early open-source LLM (3B & 7B) for general text generation; a demonstration model with basic instruction-following ability.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Command-(R,A)** (Cohere)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>CohereLabs/c4ai-command-r-v01</code>, <code>CohereLabs/c4ai-command-r7b-12-2024</code>, <code>CohereLabs/c4ai-command-a-03-2025</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Cohere’s open conversational LLM (Command series) optimized for long context, retrieval-augmented generation, and tool use.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**DBRX** (Databricks)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`databricks/dbrx-instruct`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Databricks’ 132B-parameter MoE model (36B active) trained on 12T tokens; competes with GPT-3.5 quality as a fully open foundation model.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Grok** (xAI)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`xai-org/grok-1`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>xAI’s grok-1 model known for vast size(314B parameters) and high quality; integrated in SGLang for high-performance inference.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**ChatGLM** (GLM-130B family)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`THUDM/chatglm2-6b`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Zhipu AI’s bilingual chat model (6B) excelling at Chinese-English dialogue; fine-tuned for conversational quality and alignment.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**InternLM 2** (7B, 20B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`internlm/internlm2-7b`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Next-gen InternLM (7B and 20B) from SenseTime, offering strong reasoning and ultra-long context support (up to 200K tokens).</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**ExaONE 3** (Korean-English)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`LGAI-EXAONE/EXAONE-3.5-7.8B-Instruct`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>LG AI Research’s Korean-English model (7.8B) trained on 8T tokens; provides high-quality bilingual understanding and generation.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Baichuan 2** (7B, 13B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`baichuan-inc/Baichuan2-13B-Chat`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>BaichuanAI’s second-generation Chinese-English LLM (7B/13B) with improved performance and an open commercial license.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**XVERSE** (MoE)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`xverse/XVERSE-MoE-A36B`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Yuanxiang’s open MoE LLM (XVERSE-MoE-A36B: 255B total, 36B active) supporting ~40 languages; delivers 100B+ dense-level performance via expert routing.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**SmolLM** (135M–1.7B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`HuggingFaceTB/SmolLM-1.7B`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Hugging Face’s ultra-small LLM series (135M–1.7B params) offering surprisingly strong results, enabling advanced AI on mobile/edge devices.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**GLM-4** (Multilingual 9B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`ZhipuAI/glm-4-9b-chat`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Zhipu’s GLM-4 series (up to 9B parameters) – open multilingual models with support for 1M-token context and even a 5.6B multimodal variant (Phi-4V).</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**MiMo** (7B series)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`XiaomiMiMo/MiMo-7B-RL`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Xiaomi's reasoning-optimized model series, leverages Multiple-Token Prediction for faster inference.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**ERNIE-4.5** (4.5, 4.5MoE series)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`baidu/ERNIE-4.5-21B-A3B-PT`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Baidu's ERNIE-4.5 series which consists of MoE with 47B and 3B active parameters, with the largest model having 424B total parameters, as well as a 0.3B dense model.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Arcee AFM-4.5B**</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`arcee-ai/AFM-4.5B-Base`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Arcee's foundational model series for real world reliability and edge deployments.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Persimmon** (8B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`adept/persimmon-8b-chat`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Adept’s open 8B model with a 16K context window and fast inference; trained for broad usability and licensed under Apache 2.0.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Solar** (10.7B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`upstage/SOLAR-10.7B-Instruct-v1.0`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Upstage's 10.7B parameter model, optimized for instruction-following tasks. This architecture incorporates a depth-up scaling methodology, enhancing model performance.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Tele FLM** (52B-1T)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`CofeAI/Tele-FLM`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>BAAI & TeleAI's multilingual model, available in 52-billion and 1-trillion parameter variants. It is a decoder-only transformer trained on ~2T tokens</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Ling** (16.8B–290B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>inclusionAI/Ling-lite</code>, <code>inclusionAI/Ling-plus</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>InclusionAI’s open MoE models. Ling-Lite has 16.8B total / 2.75B active parameters, and Ling-Plus has 290B total / 28.8B active parameters. They are designed for high performance on NLP and complex reasoning tasks.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Granite 3.0, 3.1** (IBM)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`ibm-granite/granite-3.1-8b-instruct`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>IBM's open dense foundation models optimized for reasoning, code, and business AI use cases. Integrated with Red Hat and watsonx systems.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Granite 3.0 MoE** (IBM)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`ibm-granite/granite-3.0-3b-a800m-instruct`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>IBM’s Mixture-of-Experts models offering strong performance with cost-efficiency. MoE expert routing designed for enterprise deployment at scale.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**GPT-J** (6B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`EleutherAI/gpt-j-6b`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>EleutherAI's GPT-2-like causal language model (6B) trained on the <a href="https://pile.eleuther.ai/">Pile</a> dataset.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Orion** (14B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`OrionStarAI/Orion-14B-Base`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>A series of open-source multilingual large language models by OrionStarAI, pretrained on a 2.5T token multilingual corpus including Chinese, English, Japanese, Korean, etc, and it exhibits superior performance in these languages.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Llama Nemotron Super** (v1, v1.5, NVIDIA)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>nvidia/Llama-3_3-Nemotron-Super-49B-v1</code>, <code>nvidia/Llama-3_3-Nemotron-Super-49B-v1_5</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>The <a href="https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/">NVIDIA Nemotron</a> family of multimodal models provides state-of-the-art reasoning models specifically designed for enterprise-ready AI agents.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**Llama Nemotron Ultra** (v1, NVIDIA)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`nvidia/Llama-3_1-Nemotron-Ultra-253B-v1`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>The <a href="https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/">NVIDIA Nemotron</a> family of multimodal models provides state-of-the-art reasoning models specifically designed for enterprise-ready AI agents.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**NVIDIA Nemotron Nano 2.0**</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>`nvidia/NVIDIA-Nemotron-Nano-9B-v2`</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>The <a href="https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/">NVIDIA Nemotron</a> family of multimodal models provides state-of-the-art reasoning models specifically designed for enterprise-ready AI agents. <code>Nemotron-Nano-9B-v2</code> is a hybrid Mamba-Transformer language model designed to increase throughput for reasoning workloads while achieving state-of-the-art accuracy compared to similarly-sized models.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>NVIDIA Nemotron 3 Super</strong> (NVIDIA)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>The <a href="https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/">NVIDIA Nemotron</a> 3 Super is a 120B-parameter MoE model (12B active) delivering high-quality reasoning and generation for enterprise AI agents.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>NVIDIA Nemotron 3 Nano</strong> (NVIDIA)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>The <a href="https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/">NVIDIA Nemotron</a> 3 Nano is a compact model designed for efficient edge and enterprise deployment with strong reasoning capabilities.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>StarCoder2</strong> (3B-15B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>bigcode/starcoder2-7b</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>StarCoder2 is a family of open large language models (LLMs) specialized for code generation and understanding. It is the successor to StarCoder, jointly developed by the BigCode project (a collaboration between Hugging Face, ServiceNow Research, and other contributors).</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>Jet-Nemotron</strong></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>jet-ai/Jet-Nemotron-2B</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Jet-Nemotron is a new family of hybrid-architecture language models that surpass state-of-the-art open-source full-attention language models, while achieving significant efficiency gains.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>Trinity</strong> (Nano, Mini)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>arcee-ai/Trinity-Mini</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Arcee's foundational MoE Trinity family of models, open weights under Apache 2.0.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>LFM2</strong> (230M, 350M, 1.2B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>LiquidAI/LFM2.5-1.2B-Instruct</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Liquid AI's hybrid language model combining gated short convolutions with a small number of grouped query attention (GQA) blocks.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>LFM2-MoE</strong> (8B-A1B, 24B-A2B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>LiquidAI/LFM2-8B-A1B</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Liquid AI's sparse Mixture-of-Experts variant of the LFM2 hybrid backbone, with SwiGLU experts, normalized sigmoid routing, and top-k expert selection.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>Falcon-H1</strong> (0.5B–34B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>tiiuae/Falcon-H1-34B-Instruct</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>TII's hybrid Mamba-Transformer architecture combining attention and state-space models for efficient long-context inference.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>Hunyuan-Large</strong> (389B, MoE)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>tencent/Tencent-Hunyuan-Large</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Tencent's open-source MoE model with 389B total / 52B active parameters, featuring Cross-Layer Attention (CLA) for improved efficiency.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>**IBM Granite 4.0 (Hybrid, Dense)**</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>ibm-granite/granite-4.0-h-micro</code>, <code>ibm-granite/granite-4.0-micro</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>IBM Granite 4.0 micro models: hybrid Mamba–MoE (<code>h-micro</code>) and dense (<code>micro</code>) variants. Enterprise-focused reasoning models</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>Sarvam 2</strong> (30B-A2B, 105B-A10B)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>sarvamai/sarvam-2</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Sarvam's Mixture-of-Experts models. The 105B variant uses MLA (Multi-head Latent Attention) and the 30B variant uses GQA, both with 128 routed experts.</td>
|
||
</tr>
|
||
<tr>
|
||
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}><strong>Laguna XS.2</strong> (poolside)</td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>poolside/Laguna-XS.2</code></td>
|
||
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>Poolside's hybrid sliding-window-attention MoE model (256 routed experts, 1 shared expert, sigmoid router) with per-layer-type RoPE (YARN on full-attention layers, default on sliding-attention layers) and a softplus per-head attention gate.</td>
|
||
</tr>
|
||
</tbody>
|
||
</table>
|