Filter by type:
Filter by tags:
deepseek-v4-flash
# DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
Technical Report👁️
## Introduction
We present a preview version of **DeepSeek-V4** series, including two strong Mixture-of-Experts (MoE) language models — **DeepSeek-V4-Pro** with 1.6T parameters (49B activated) and **DeepSeek-V4-Flash** with 284B parameters (13B activated) — both supporting a context length of **one million tokens**.
DeepSeek-V4 series incorporate several key upgrades in architecture and optimization:
1. **Hybrid Attention Architecture:** We design a hybrid attention mechanism combining Compressed Sparse Attention (CSA) and Heavily Compressed Attention (HCA) to dramatically improve long-context efficiency. In the 1M-token context setting, DeepSeek-V4-Pro requires only **27% of single-token inference FLOPs** and **10% of KV cache** compared with DeepSeek-V3.2.
2. **Manifold-Constrained Hyper-Connections (mHC):** We incorporate mHC to strengthen conventional residual connections, enhancing stability of signal propagation across layers while preserving model expressivity.
3. **Muon Optimizer:** We employ the Muon optimizer for faster convergence and greater training stability.
...
Repository: localaiLicense: mit
laguna-xs-2.1
Laguna XS 2.1 is Poolside's 33B-parameter, 3B-active Mixture-of-Experts model for agentic coding and long-horizon work on local machines. It supports tool use, interleaved reasoning, and a native 262K-token context window. This default entry uses the official 20.3 GB Q4_K_M GGUF.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-xs-2.1-apex-i-quality
Laguna XS 2.1 in the 21.8 GB APEX-I Quality format. This is the highest-fidelity importance-matrix APEX build for llama.cpp.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-xs-2.1-apex-i-balanced
Laguna XS 2.1 in the 24.3 GB APEX-I Balanced format, an importance-matrix build balancing fidelity and memory use for llama.cpp.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-xs-2.1-apex-i-compact
Laguna XS 2.1 in the 15.8 GB APEX-I Compact format, an importance-matrix build tuned for lower memory use in llama.cpp.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-xs-2.1-apex-i-mini
Laguna XS 2.1 in the 12.8 GB APEX-I Mini format, the smallest importance-matrix APEX build for llama.cpp.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-xs-2.1-apex-quality
Laguna XS 2.1 in the 21.8 GB APEX Quality format, the highest-fidelity non-imatrix APEX build for llama.cpp.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-xs-2.1-apex-balanced
Laguna XS 2.1 in the 24.3 GB APEX Balanced format, balancing fidelity and memory use for llama.cpp.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-xs-2.1-apex-compact
Laguna XS 2.1 in the 15.8 GB APEX Compact format, tuned for lower memory use in llama.cpp.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-s-2.1-q8
Laguna S 2.1 is Poolside's 118B-parameter, 8B-active Mixture-of-Experts model for agentic software engineering. It supports tool use and a native one-million-token context window; the official GGUF recommends 256K context for best output quality. This entry uses the 129 GB Q8_0 build, with routed experts quantized to Q8_0 and the signal path kept in BF16.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-s-2.1
Laguna S 2.1 is Poolside's 118B-parameter, 8B-active Mixture-of-Experts model for agentic software engineering. It supports tool use and a native one-million-token context window; the official GGUF recommends 256K context for best output quality. This default entry uses the current 96 GB Q4_K_M artifact, with imatrix-quantized routed experts and a Q8_0 signal path.
License: OpenMDW 1.1.
Repository: localaiLicense: other
laguna-s-2.1-dflash
Laguna S 2.1's 96 GB Q4_K_M target paired with Poolside's 2.2 GB BF16 DFlash drafter for speculative decoding. DFlash drafts a block of tokens per forward pass and accelerates generation without changing the target model's outputs. Flash attention is enabled because the DFlash path requires it.
License: OpenMDW 1.1.
Repository: localaiLicense: other
ai21labs_ai21-jamba-reasoning-3b
AI21’s Jamba Reasoning 3B is a top-performing reasoning model that packs leading scores on intelligence benchmarks and highly-efficient processing into a compact 3B build.
The hybrid design combines Transformer attention with Mamba (a state-space model). Mamba layers are more efficient for sequence processing, while attention layers capture complex dependencies. This mix reduces memory overhead, improves throughput, and makes the model run smoothly on laptops, GPUs, and even mobile devices, while maintainig impressive quality.
Repository: localaiLicense: apache-2.0
ibm-granite_granite-4.0-h-small
Granite-4.0-H-Small is a 32B parameter long-context instruct model finetuned from Granite-4.0-H-Small-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging. Granite 4.0 instruct models feature improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
Repository: localaiLicense: apache-2.0
ibm-granite_granite-4.0-h-tiny
Granite-4.0-H-Tiny is a 7B parameter long-context instruct model finetuned from Granite-4.0-H-Tiny-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging. Granite 4.0 instruct models feature improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
Repository: localaiLicense: apache-2.0
ibm-granite_granite-4.0-h-micro
Granite-4.0-H-Micro is a 3B parameter long-context instruct model finetuned from Granite-4.0-H-Micro-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging. Granite 4.0 instruct models feature improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
Repository: localaiLicense: apache-2.0
ibm-granite_granite-4.0-micro
Granite-4.0-Micro is a 3B parameter long-context instruct model finetuned from Granite-4.0-Micro-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging. Granite 4.0 instruct models feature improved instruction following (IF) and tool-calling capabilities, making them more effective in enterprise applications.
Repository: localaiLicense: apache-2.0
baidu_ernie-4.5-21b-a3b-thinking
Over the past three months, we have continued to scale the thinking capability of ERNIE-4.5-21B-A3B, improving both the quality and depth of reasoning, thereby advancing the competitiveness of ERNIE lightweight models in complex reasoning tasks. We are pleased to introduce ERNIE-4.5-21B-A3B-Thinking, featuring the following key enhancements:
Significantly improved performance on reasoning tasks, including logical reasoning, mathematics, science, coding, text generation, and academic benchmarks that typically require human expertise.
Efficient tool usage capabilities.
Enhanced 128K long-context understanding capabilities.
Note: This version has an increased thinking length. We strongly recommend its use in highly complex reasoning tasks. ERNIE-4.5-21B-A3B-Thinking is a text MoE post-trained model, with 21B total parameters and 3B activated parameters for each token.
Repository: localaiLicense: apache-2.0
huggingfacetb_smollm3-3b
SmolLM3 is a 3B parameter language model designed to push the boundaries of small models. It supports 6 languages, advanced reasoning and long context. SmolLM3 is a fully open model that offers strong performance at the 3B–4B scale.
The model is a decoder-only transformer using GQA and NoPE (with 3:1 ratio), it was pretrained on 11.2T tokens with a staged curriculum of web, code, math and reasoning data. Post-training included midtraining on 140B reasoning tokens followed by supervised fine-tuning and alignment via Anchored Preference Optimization (APO).
Repository: localaiLicense: apache-2.0
qwen3-235b-a22b-instruct-2507
We introduce the updated version of the Qwen3-235B-A22B non-thinking mode, named Qwen3-235B-A22B-Instruct-2507, featuring the following key enhancements:
Significant improvements in general capabilities, including instruction following, logical reasoning, text comprehension, mathematics, science, coding and tool usage.
Substantial gains in long-tail knowledge coverage across multiple languages.
Markedly better alignment with user preferences in subjective and open-ended tasks, enabling more helpful responses and higher-quality text generation.
Enhanced capabilities in 256K long-context understanding.
Repository: localaiLicense: apache-2.0
menlo_lucy-128k
Lucy is a compact but capable 1.7B model focused on agentic web search and lightweight browsing. Built on Qwen3-1.7B, Lucy inherits deep research capabilities from larger models while being optimized to run efficiently on mobile devices, even with CPU-only configurations.
We achieved this through machine-generated task vectors that optimize thinking processes, smooth reward functions across multiple categories, and pure reinforcement learning without any supervised fine-tuning.
Repository: localaiLicense: apache-2.0
Page 1