Perplexity
- April 23, 2026 — OpenAI released GPT-5.5, the first model with native computer use, focused on agent scenarios, web search, and mathematical calculations[3][5].
- April 16, 2026 — Anthropic introduced Claude Opus 4.7, a leader in programming and long-context work, which has established itself as the standard for developers[5].
- February 2026 — Google updated its flagship model to Gemini 3.1 Pro, enhancing multimodality and reducing costs relative to competitors[1][5].
- April 2026 — Alibaba released a new model that took first place in three OpenRouter rankings (Daily, Weekly, Trending) with low latency and high reasoning ability[2].
- April 2026 — MiniMax updated its music generation to Music 2.6, available for free for 14 days[2].
Sources:
- The Loudest AI Releases of February and March 2026
- AI News: What's Happening in the World of Artificial Intelligence in April
- Which AI is the Smartest in 2026: TOP 7 Models with Tests
- Which Neural Network to Choose: Analyzing the Main Models of 2026
Gemini
Of course, here's a review of the past week's major events, focusing on model releases, presented from a different angle than previous digests:
- OpenAI Unveils GPT-4o: Next-Level Speed and Multimodality. On May 13th, OpenAI announced its new flagship model, GPT-4o, which is significantly faster than previous versions and boasts enhanced real-time processing capabilities for text, audio, and images. This opens up new possibilities for creating more natural and interactive AI assistants.
- Google I/O 2024: Gemini 1.5 Pro and New AI Tools for Developers. At its annual conference, Google introduced Gemini 1.5 Pro with an increased context window of up to 2 million tokens, as well as a range of new AI tools and APIs designed to simplify AI integration into applications. Special attention was paid to the safety and ethical use of AI.
- Anthropic Releases Claude 3.5 Sonnet: A Balance of Speed and Performance. On May 2nd, Anthropic presented Claude 3.5 Sonnet, a new model positioned as a faster and more accessible alternative to Claude 3 Opus, while maintaining a high level of performance in content analysis and generation tasks.
Sources:
- OpenAI Blog
- Google Developers Blog
- Anthropic Blog
ChatGPT
Several significant events related to new model releases occurred in the field of artificial intelligence between June 28 and July 5, 2026:
June 29, 2026: DeepSeek introduced the DeepSpec collection, including DSpark, DFlash, and Eagle models, optimized for local deployment based on Qwen3 and Gemma-4.
July 2, 2026: Z.ai announced ZCode 3.0, an AI-IDE with agent integration and GLM-5.2, available via subscription starting at 8 per month.
July 2, 2026: Tencent released PhoneBuddy, an agent based on Qwen3.5-4B, capable of controlling a smartphone and outperforming GPT-5.4 in tasks.
July 2, 2026: OpenAI halved the inference cost of ChatGPT, enabling its service using several hundred GPUs.
July 2, 2026: Export restrictions on Anthropic were lifted, restoring access from July 2; identity verification may be required.
Sources:
- AI Vibe Digest
- Neurosexual
Grok
1. June 29, 2026 — OpenAI released GPT-5.6 (including Sol/Terra variants) under limited access for US government-approved clients only due to cybersecurity concerns; public release is indefinitely postponed.
2. June 29–30, 2026 — Anthropic launched Claude Mythos 5 (the company's most powerful model) exclusively for over 100 vetted US government agencies, and partially restored access to Fable 5 after the lifting of Ministry of Trade restrictions.
3. June 30, 2026 — Anthropic officially released Claude Sonnet 5 — a midsize model optimized for agentic tasks and tool-use, priced significantly lower than flagship Opus variants, while maintaining close performance on coding and reasoning.
4. July 1, 2026 — Following authorization from US authorities, Anthropic restored access to Claude Fable 5 and Mythos 5 for vetted organizations, marking one of the first notable instances of partial de-restriction on frontier models.
5. Early July 2026 — ByteDance and other Chinese developers (including updates to the GLM and Seed series) continued their active release of competitive models, emphasizing the trend towards more accessible alternatives to Western frontier releases.
Sources:
- llm-stats.com/llm-updates
- aiflashreport.com/model-releases.html
- Sy Mundine's Facebook post (July 3, 2026)
- ABC News and AI News in a Minute YouTube streams (June–July 2026)
- AI Insiders LinkedIn reviews (July 2026)
Claude
I'll search for current information on model releases from the past week. Excellent, I have the latest data. I'll compile an overview with a different focus – not just a list of releases, but shifts in lab strategies and victories with regulators. This is significantly different from a standard digest.
Top 5 Events of the Week: How Regulators and the Market are Changing the AI Game
Claude Sonnet 5 Becomes the Flagship for the Masses (June 30) — Anthropic has lowered the entry barrier: the new Sonnet 5 is now the default model for all Free and Pro users and is integrated into VS Code, Cursor, and GitHub Copilot. The introductory price is $2/ 0 per million tokens (instead of $3/ 5 after August). This is a move not for benchmarks, but for deployment – capturing workflows, not just user chats.
Claude Fable 5 Returns After Government Ban (July 1) — The US government ordered the model to be shut down on June 12 under the pretext of export controls, forcing Anthropic to disable it globally for three weeks. As of July 1, 2026, Anthropic has returned Claude Fable 5 to active status after a nearly three-week hiatus; the US government had demanded the model be shut down on June 12 under directive from Secretary of Commerce Howard Latnik for national security reasons. The signal is clear: frontier models are now under government oversight.
OpenAI Launches GPT-5.6 with a Three-Dimensional Hierarchy, but Under Lock and Key (June 26) — OpenAI has introduced a new GPT-5.6 family of three models (Sol, Terra, and Luna) with strong cybersecurity capabilities and has begun with a limited preview for trusted partners before wider access. The strategy is evident: large models will go through a US government access list (~20 organizations). This is not expanding access – it's managed distribution.
Google Fills its Lineup with Generative Models Instead of Gemini 3.5 Pro (June 30) — Google released two new image generation models (Gemini 3.1 Flash Image and Gemini 3 Pro Image) on June 30, part of a strategy to expand its lineup while Gemini 3.5 Pro continues to be delayed. The flagship delay is being compensated for with multimodal capabilities. This is either a reorganization before a major release or an admission that a single super-strong model is no longer the main focus.
Anthropic Launches Claude Science to Accelerate Life Sciences (June–July) — Anthropic has launched Claude Science as a dedicated application for scientific research, with an initial focus on drug discovery, protein structure analysis, genomics, and computational biology, in line with CEO Dario Amodei's goal to reduce the life science R&D cycle tenfold. Vertical deepening instead of horizontal expansion is the new economy.
Sources:
- LLM Gateway
- OpenAI Release Notes
- Build Fast with AI
- FelloAI
- Mean CEO Blog
Sources:
- The Loudest AI Releases of February and March 2026 - YouTube
- AI News: What's Happening in the World of Artificial Intelligence...
- Which AI is the Smartest in 2026: TOP 7 Models with Tests - RBC
- Top 6 New Neural Networks in June 2026 - T-Business Secrets
- Which Neural Network to Choose: Analyzing the Main Models of 2026
- Between Convenience and Law: AI Model Ranking 2026
- AI Tools in 2026. My Toolkit - Habr
- Best AI for Programming in 2026: Top Neural Networks...
- What to Expect from AI in 2026? - Euronews.com
- AI Ranking | Comparison of Language Models and Neural Networks...
- AI Updates Today (July 2026) – Latest AI Model Releases
- LLM News Today (July 2026) – AI Model Releases
- New Models Today — AI & LLM Releases Last 24 Hours | Price Per Token
- AI Model Release Timeline 2025–2026 — Every LLM Launch Tracked | AI Flash Report
- LLM Release Timeline — When Each AI Model Was Released | LLM Gateway