#Large Language Models
AIKimi K3 Is the World's First Open 3T Model - Beats GPT-5.6 Sol on Coding at $3 Per Million Tokens
Kimi K3 is live today - Moonshot's 2.8T open model beats GPT-5.6 Sol on GPU kernel work, offers a $3/MTok API, and drops full weights July 27.
AIGoogle's Gemini 3.5 Pro Is Months Late - Coding Benchmarks Fell Short
Gemini 3.5 Pro is months behind schedule after Google's coding benchmark push fell short, leaving only 3.1 Pro available as OpenAI and Meta advance.
AIAI Models Refuse Criticism of Repressive Governments Twice as Often, Study Finds
AI models refused political criticism of China and Saudi Arabia 34% of the time - double the rate for permissive regimes. Meta's Oversight Board tested 10 models in its first LLM study.
AIDeepSeek Raised $7.4 Billion Last Month at $50 Billion. Now It Wants $1.5 Billion More and a 2027 IPO
DeepSeek raised $7.4B in June at a $50B valuation. Bloomberg reports it's already in talks for another $1.5B at $71B - and targeting a 2027 IPO.
AIOpenAI Cut GPT-5.6 Sol's Thinking Budget 87% - Four Days After Launch
OpenAI cut the GPT-5.6 Sol thinking budget across all tiers four days after launch. Sol max fell from 960 to 128. Terra and Luna were not affected.
AIAI API Pricing Compared 2026: What Every Major Model Costs per Token
Output tokens cost anywhere from $0.20 to $50 per million depending on the lab. Every major AI API rate in one table, verified July 10, 2026.
AIOpenAI's GPT-5.6 Is Now the Preferred Model in Microsoft 365 Copilot
GPT-5.6 is now the default in Microsoft 365 Copilot - Word, Excel, PowerPoint, and Chat. The catch: Microsoft is also quietly building its own AI models.
AIMeta Opens Its First Developer API With Muse Spark 1.1
Meta Model API opens in public preview today. Developers get Muse Spark 1.1 - Meta's agentic model with 1M context and an OpenAI-compatible endpoint.
AIGPT-5.6 Sol Goes Public After 12 Days Behind a White House Gate
OpenAI's GPT-5.6 family goes public today after 12 days behind a White House gate. Sol, Terra, and Luna are now live - and Sol gamed its own safety tests.
AIOllama Raises $65M Series B, Now at 8.9 Million Developers
Ollama raised $65M Series B, led by Theory Ventures and Benchmark. The platform counts 8.9 million developers and adds nearly 1M installs weekly.
AIGrok 4.5 vs Claude Fable 5 vs GPT-5.5: Benchmarks, Pricing, and Who Wins for Coding
Grok 4.5 vs Claude Fable 5 vs GPT-5.5: benchmark scores, pricing per million tokens, and token efficiency compared across four coding evaluations.
AIOpenAI Launches GPT-Live - a Full-Duplex Voice Model That Talks While It Listens
GPT-Live launches today across ChatGPT globally. Full-duplex audio lets it listen and speak at once - delegating searches and reasoning to GPT-5.5 in the background.