Videos

Meta's Muse Wonder models are competitive frontier models that score number one on the Deep Suite coding benchmark...

Meta's Muse Wonder models benchmark extremely high, but in hands-on coding and 3D game tests they underperform...

Claude Haiku 5.5 is priced at 10 cents per million input tokens and 50 cents per million output tokens for prompts...

OpenAI's September 29th halving of ChatGPT Pro allowances means 19 months of the $200 plan now costs the same as a...

Sonnet 5.5 performs nearly identically to Opus 5.5 on coding and reasoning benchmarks while costing half as much —...

Claude Sonnet 5.5 delivers a massive jump in agentic coding capability, scoring 70.6% compared to Sonnet 5's 10.3%,...

Space Bunny Alpha is a new stealth model on Open Router that appears to be a competent flash-level model with fast...

Mimo V2.6 Flash, the smaller 309B model, proved more reliable and capable than the trillion-parameter Pro in...

MIT's superposition analysis shows that LLM loss scales as roughly 1/width, meaning doubling a model's width halves...

Opus 5.5 won 7 of 10 use cases against GPT-6 Soul in this head-to-head test, consistently producing better creative...

Grok 4.7 is a cost-effective model that performs near the frontier, comparable to Opus 5 on Cursorbench at half the...

Cognition's SWE-2 is a post-trained version of Kimi K3 that improves agentic coding, matching GPT-4.5 on DeepSWE...

Gemini 3.8 Flash is a fast, cheap model that shows clear improvement over 3.7 Flash, especially in a robotic arm...

Fable 5 built a comparable app to Fable 5.1 for $500 and half a day, while Fable 5.1 cost $1200 and took 1.5 days....

Qwen3.8 Flash Next, a preview of the Qwen 4 architecture, demonstrates exceptional vision and code generation...

DeepSeek V4 Flash Vision (experimental) adds vision to a small model, and the results on coding tasks like browser...

Deepseek Harness is an open-source agent harness that offers unmatched customization and speed compared to Claude...

Ox Alpha is a mysterious new reasoning model available on OpenRouter, and based on Bijan Bowen's tests, it produces...

Google's Gemini 3.7 Flash is here, and it's faster, cheaper, and more capable than its predecessors. In tests, it...

GLM 5.3 is a 753B-parameter open-weight mixture-of-experts model that matches closed-source giants like Kimi K3 on...

ChatGPT just got 14x faster thanks to Cerebras custom chips, making the model no longer the bottleneck—your computer...

Qwen 3.8 27B is a powerful local AI model that can run on a single GPU, showing impressive results in coding, game...

Qwen 3.8 27B is a huge leap over 3.6 for local AI agentic coding. In a zero-shot test, it built three playable retro...

Google's Gemini 3.7 Flash beats Claude Sonnet 5 on coding benchmarks at $0.75 per million tokens — half the price of...

Grok 4.6 ties OpenAI's top model on a key benchmark at a fifth of the output price, but the model is the least...

Grok 4.6 is a serious competitor to models from OpenAI and Anthropic, especially in coding and front-end tasks. The...

DeepSeek V4 Pro (0813) is out of preview and it's a big leap over the undercooked preview version, but it's not the...

Meta released Muse Glimmer, a 30B-parameter open-source model (Apache 2.0) designed to fit on a 24GB GPU, built via...

Ling 3.0 Tiny, a 7.9B parameter mixture-of-experts model with 1.3B active parameters, impresses in coding tests...

Alibaba released Qwen3.8 Max, a 2.4 trillion parameter mixture-of-experts model with 95 billion active parameters,...

David Ondrej open-sourced his personal repository of 42 agent skills for AI coding tools like Codex and Claude Code,...

OpenAI's GPT-5.6 Luna is now its cheapest model after an 80% price cut, costing $0.20 per million input tokens and...

DeepSeek V4 Flash is a newly released official version of a small but powerful AI model with 284B parameters (13B...

Bottle Cap AI's ThinkingCap fine-tune of Qwen 3.6 27B reduces reasoning tokens by ~46% while preserving benchmark...

Ling 3.0 Flash from Ant is a surprisingly competent coding model at 124B total/5.1B active parameters, outperforming...

DeepSWE is a contamination-resistant coding benchmark from Datacurve, composed of 113 original software engineering...

Claude Opus 5 is often cheaper than Fable 5 and can outperform it on coding and verification tasks, but Fable 5...

Anthropic's Claude Opus 5 was released, outperforming the larger Fable 5 on most benchmarks while costing about half...

Claude Opus 5 has been released with state-of-the-art performance on coding and knowledge work benchmarks, often...

Kimi K3 is the most powerful open-weight model released, but it suffers from reliability issues that public...

Poolside Laguna S2.1 is a 118B parameter Mixture of Experts model (8B active) that excels at creative writing and...

Kimi K3 is an open-source AI model from Moonshot AI that matches or beats closed-source models like Fable 5 and...

Poolside's co-founder Eiso Kant argues that model building is primarily an engineering discipline, and that their...

Google's Gemini 3.6 Flash is a new model that offers significant speed improvements, a 17% reduction in token usage,...

Kimi K3 is a massive 2.8 trillion parameter open-weight model that claims to rival proprietary models like Claude...

Meta's Muse Spark 1.1 shows benchmark improvements over its predecessor but fails in practical coding and game...

GPT 5.6 Soul outperforms Fable 5 in three out of four real-world coding and design tasks at a fraction of the cost,...

OpenAI released GPT-5.6 as a trio of models—Soul, Terra, and Luna—each with different cost and capability tiers,...

Grok 4.5 is a 1.5-trillion-parameter frontier model from xAI, built on a new V9 foundation that incorporates...

Grok 4.5 is a highly competitive AI model that rivals state-of-the-art models like GPT-5 and Opus 4, offering strong...