Live19 today

Tags

Models

24

Models19h ago

Cognition launches a cheaper, faster coding model — but keeps the weights private

SWE-2 Flash runs inside Devin at one-fifth the price and three times the speed of SWE-2, while cyber-risk concerns limit access to its weights.

Models21h ago

webAI opens a local 8.8B reasoning model — but bans commercial use

Models21h ago

Ricursive opens Ripple-65B after training it for $32 million

Models1d ago

Moonshot opens Kimi K3.1 Pro weights for self-hosting

The 2.8-trillion-parameter multimodal model has 104 billion active parameters, a million-token context window and commercial-use conditions.

Models1d ago

Pruna turns MiniMax H3 into a 5x faster open-weight image model

P-H3-Image generates and edits 2–16 MP images, while its smallest build is a 20 GB download and uses about 21 GB of VRAM at 8 MP.

More stories

Models1d ago

OpenRouter offers a free 1M-context multimodal model for agent work

Crouton Alpha accepts text, images, audio and video, but its developer remains undisclosed and access is temporary and region-limited.

Models1d ago

Tencent launches Hy4 with vision and computer use

The low-cost model works across GUI and CLI tasks, while open weights and the full system card are still promised rather than available.

Models6d ago

Pika releases open-weight video model with synchronized audio

H3 Flex runs locally as a six-step LoRA on MiniMax H3, while Pika offers hosted 4K upscaling and separate audio stems.

ModelsOct 29, 2026

Mistral opens its 1-trillion-parameter model for coding and cyber defense

Mistral Large 4, nicknamed Le Chonk, can now be downloaded, customized and self-hosted under a modified MIT license.

ModelsOct 29, 2026

DeepSeek and Moonshot reportedly trained new models on GPT-6 Astra outputs

The Information says Kimi K3.1 Pro and DeepSeek V4.1-Flash benefited from distillation of OpenAI’s frontier model; the companies did not immediately comment.

ModelsOct 28, 2026

Cartesia says H-Net matches a 3B model with 10x less training

The tokenizer-free architecture learns from raw bytes and reaches SmolLM3-level results with roughly one-tenth the data and compute, according to Cartesia.

ModelsOct 28, 2026

Particle launches a coding model that uses 45% fewer tokens

Particle-1 post-trains DeepSeek V4.1 Flash for software agents, promising comparable task accuracy with lower cost and latency; the benchmarks are the company’s own.

ModelsOct 28, 2026

Google launches Gemini 4 Carbon at the top of the AI intelligence index

The frontier model improves cyber and long-horizon work, but access starts with selected defenders before a broader rollout.

ModelsOct 26, 2026

OpenAI launches GPT-6.1 Terra for cheaper computer-use agents

The new model targets everyday computer work with lower costs while extending OpenAI’s GPT-6.1 lineup across ChatGPT, Codex and the API.

ModelsOct 17, 2026

xAI launches Grok Voice 3.0 for faster, cheaper voice agents

The speech-to-speech model leads Artificial Analysis's index, responds in under a second and keeps the previous version's API price.

ModelsOct 12, 2026

Meta launches Muse Wonder as a frontier model for agents

The new Muse Wonder and Muse Wonder Full models target coding, long-running tasks and professional domains, with access through Meta's model API and Muse Code.

FundingOct 10, 2026

Nvidia weighs buying Reflection AI to strengthen open-weight models

Financial Times reports early talks around an acquisition, deeper investment, or acqui-hire of the startup behind Beam.

ModelsOct 7, 2026

OpenAI rolls out GPT-6 with interactive answers in ChatGPT

The new default ChatGPT model can stream text, visuals, controls, charts and small tools into the conversation, with Free access following paid tiers.

ModelsOct 4, 2026

Reflection prepares open-weight model to challenge Chinese AI labs

Axios says the Nvidia-backed startup is getting ready to release an American open-weight system that could give companies a cheaper alternative to closed frontier models.

ModelsOct 3, 2026

OpenAI explains when to use each GPT-6 model

The guide splits the family into Astra for hardest reasoning, Sol for complex coding and research, and Luna for focused production tasks at scale.

ModelsSep 30, 2026

Google launches Gemini 4 Argon for cyber defense and complex work

The frontier model is starting with trusted cyber defenders before broader API, enterprise and consumer access.

ModelsSep 29, 2026

OpenAI launches GPT-6.1 Sol and always-on dots agents

Sol brings near-Astra coding and professional-work performance at much lower API prices, while dots are proactive agents that can work across apps with approvals.

ModelsSep 10, 2026

DeepSeek pushes AI model prices lower with V4.1 Flash

The new API model supports a 1M-token context, tool calls and OpenAI- or Anthropic-compatible access, with cache-hit input priced from $0.003 per 1M tokens.

Aug 22, 2026

OpenAI cuts GPT-5.6 Sol API prices for three months

The frontier model now costs $4 per 1M input tokens and $20 per 1M output tokens for standard short-context API use.

End of feed