Search results for «MoE Architecture»
-
Clover-35B-A3B-Preview Review: DiffuSpace Open-Weight MoE Model Capabilities and Access Options
Clover-35B-A3B-Preview is DiffuSpace's first open-weight block-diffusion MoE model derived from Qwen3.6-35B-A3B with 35 billion total parameters.
-
Mistral Large 4 Review: Open-Weight 1T-Parameter MoE Model, API Access and Cybersecurity Capabilities
Mistral Large 4 is a 1T-parameter multimodal MoE model with open weights, strong cybersecurity benchmarks and API access at roughly $0.68 per million input tokens.
-
Kimi K3.1 Pro Review: Moonshot AI 1M Context Model, API Pricing and Highspeed Variant
Kimi K3.1 Pro launched October 23, 2026 as a MoE model with native 1M token context, tool calling, and two API tiers priced at $2/$10 and $4/$20 per million tokens.
-
Kimi K3.1 Pro Review: Moonshot AI Model Capabilities and API Costs
Kimi K3.1 Pro launched October 23, 2026 as Moonshot AI's first K3.1 series model with a 1,048,576-token context window and MoE architecture.
-
Recursive Flint Review: Local Coding Model Features, Benchmarks and Deployment Options
Recursive Flint is a 39.8B-parameter open-weight MoE coding model released October 21, 2026, optimized for local agentic tasks like bug fixing and test writing.
-
MiniMax M3.1 Flash Review: Agentic Performance, Pricing and Limitations Explained
MiniMax M3.1 Flash is a 428B-parameter MoE model released October 19, 2026, for agentic workflows with 1M context and flat-rate pricing at $0.15 per million input tokens.
-
Meta Muse Spark Open Review: Open Weights Release, Multimodal Capabilities and Context Limits
Meta released Muse Spark Open on October 15, 2026, as an open-weight MoE model with 816B total parameters and a 1M-token context window for agentic and multimodal tasks.
-
DeepSeek V4.1 Flash review: Open Weights, Agentic Performance and Costs
DeepSeek V4.1 Flash 1017 offers MIT-licensed weights and improved agent scores on a 552B MoE model with 1M context, priced from $0.15 per million input tokens.
