AI Agents
AI Agent technology and applications
Articles
DeepSeek Launches V4.1-Flash-1017, Scores 77.0 on DeepSWE andNEWβ8
DeepSeek releases V4.1-Flash-1017, scoring 77.0 on DeepSWE with faster tasks and shorter outputs, plus open Search and Fetch APIs for developers.
StepFun Releases Step 5 Preview as Open Weights: 600B MoE, 27BNEWβ9
StepFun releases Step 5 Preview open weights: a 600B MoE model with 27B active parameters, 1M-token context, and strong long-context reasoning for agentic workl...
20 Jev Alternatives: The Decision Models Taking On TypeSafe in 2026NEWβ8
Compare 20 Jev alternatives from Cloudflare, OpenAI, AWS and more on size, license, pricing, API compatibility and benchmarksβplus which models beat Jev.
Perplexity Open-Sources SPACE: A Firecracker MicroVM Sandbox forNEWβ7
Perplexity open-sources SPACE, a Firecracker microVM sandbox for secure AI agents, offering production-tested isolation under the Apache 2.0 license.
NVIDIA Releases Nemotron 3.5 Super VL, a 120B Open Multimodalβ9
NVIDIA launches Nemotron 3.5 Super VL, a 120B open multimodal agent model activating 12B parameters per token, with a 1M context window for enterprise AI.
CopilotKit Launches CLiF: A Continual Learning Platform Thatβ8
CopilotKit's CLiF platform turns agent traces and user actions into skills, production evals, and fine-tuning data, now in early access.
openKylin Unveils KART Unified AI Runtime and System-Levelβ9
openKylin unveiled KART, a unified AI runtime, and system-level agent execution at Open Source Summit Europe 2026, reimagining Linux for the AI era.
TypeSafe AI Launches Jev 1.14 with Image Support, Box Type, andβ8
TypeSafe AI's Jev 1.14 adds image input, a box type, and 262k context. Explore these experimental features for typed decision models and vision tasks.
Meet Grok Voice 3.0: SpaceXAI's Voice Agent Model Ranks #1 onβ8
SpaceXAI's Grok Voice 3.0 debuts at #1 on Artificial Analysis's Speech-to-Speech Index with 86.9, offering unchanged $0.08/min pricing via the Voice Agent API.
Datalab Releases OmniParseBench: An Open OCR Benchmark With 16β9
Datalabβs OmniParseBench is an open OCR benchmark scoring parsers with 16,288 pass/fail tests across 2,937 pages, 92 languages.