|
1 | | -## יוני 2025 |
2 | | -- גוגל שחררו עדכון חדש למודל Gemini 2.5 pro, שהפך למודל SOTA על כל רחבי הלוח |
3 | | -- elevanlabs שחררה את גרסה v3 של מודל ה-tts שלה עם יכולות דיבור טבעיות במגוון שפות |
4 | | -- OpenAI o3 pro |
5 | | - |
6 | | - |
7 | | -## יולי 2025 |
8 | | -- Grok 4 שוחרר במספר דגמים שונים, והוא מציג ביצועי SOTA בכל רחבי הלוח |
9 | | -- openAI מציגה את Agent. סוכן המשולב בתוך ממשק צ'אט GPT, המשלב יכולות קידוד, מחקר מעמיק וסוכן, ומציג ביצועי SOTA במספר מדדים מורכבים |
10 | | -- מודל נסיוני של openAI זוכה במדליית זהב ב-IMO, ללא שימוש בכלים |
11 | | -- מודל מתקדם של gemini בשילוב חשיבה עמוקה, זוכה אף הוא במדליית זהב ב-IMO |
12 | | -- qwen שחררה בקוד פתוח את Qwen3-235B-A22B-Instruct-2507 ואת qwen 3 coder |
13 | | -- שוחרר מודל kimi K2 הסיני |
14 | | - |
15 | | -## July 2025 |
16 | | - |
17 | | -* xAI releases **Grok 4** in Standard, Premium+ and SuperHeavy tiers, offering a 256K-token context window and always-on reasoning. |
18 | | -* OpenAI unveils the **ChatGPT Agent**, embedding autonomous coding, web research and tool use directly within the chat interface. |
19 | | -* An experimental OpenAI model secures a **gold medal** at IMO 2025 without any external tools. |
20 | | -* Google introduces **Gemini Deep Think**, which also earns an IMO 2025 gold by solving five of six problems with parallel reasoning. |
21 | | -* Alibaba open-sources two variants—**Qwen3-235B-A22B-Instruct-2507** (instruction-tuned) and **Qwen3-Coder**—for general LLM use and automated code generation. |
22 | | -* Moonshot AI debuts **Kimi K2**, a Chinese LLM praised for its open-research focus and robust performance. |
23 | | -* Chinese startup Zhipu open-sources **GLM-4.5**, a 130 B-parameter model tailored for intelligent-agent applications. |
24 | | - |
25 | | - |
26 | 1 | ## August 2025 |
27 | 2 | - Google introduced **Gemini 2.5 Deep Think**, a special "extended thinking" mode for solving complex problems and exploring alternatives. (*special*) |
28 | 3 | - Anthropic released **Claude Opus 4.1**, an upgrade focused on improving agentic capabilities and real-world coding. |
29 | 4 | - Google DeepMind announced **Genie 3.0**, a "world model" for creating interactive 3D environments from text, maintaining consistency for several minutes. (*special*) |
30 | 5 | - OpenAI released **gpt-oss-120b** and **gpt-oss-20b**, a family of open-source models with high reasoning capabilities, optimized to run on accessible hardware. |
31 | 6 | - OpenAI launched **GPT-5**, the company's next-generation model, with significant improvements in coding and a dynamic "thinking" mode to reduce hallucinations. (*special*) |
32 | 7 | - DeepSeek released **DeepSeek V3.1**, a hybrid model combining fast and slow "thinking" modes to improve performance in agentic tasks and tool use. |
33 | | -- Google launched a preview of **Gemini 2.5 Flash Image** (showcased as *nano-banana*), an advanced model for precise image editing, merging, and maintaining character consistency. (*special*) |
| 8 | +- Google launched a preview of **Gemini 2.5 Flash Image** (showcased as *nano-banana*), an advanced model for precise image editing, merging, and maintaining character consistency. (*special*) |
| 9 | + |
| 10 | +## ספטמבר 2025 |
| 11 | +- Seedream 4.0 של ByteDance הוא מודל דור חדש ליצירת ועריכת תמונות במהירות ובאיכות גבוהה (עד 4K), המשלב יצירה ועריכה במערכת אחת ומתחרה ישירות במובילי השוק כמו Google DeepMind. |
| 12 | +- Deep Think achieves Gold Medal at the ICPC 2025 Programming Contest |
| 13 | +- OpenAI Reasoning Model Solved ALL 12 Problems at ICPC 2025 Programming Contest |
| 14 | +- suno V5 שוחרר |
| 15 | +- qwen 3 max |
| 16 | +- wan 2.2 - מודל וידאו גנרטיבי מתקדם המסוגל לערוך והלחליף דמות לאורך סרט שלם |
| 17 | +- אנתרופיק משחררת את מודל קלוד סונט 4.5 |
| 18 | +- openAI משיקה את SORA 2 |
| 19 | + |
| 20 | + |
| 21 | +## September 2025 |
| 22 | + |
| 23 | +- ByteDance released **Seedream 4.0**, a next-generation image model unifying high-quality text-to-image generation and natural-language image editing. |
| 24 | +- An advanced Gemini variant, reported as **Gemini 2.5 - Deep Think**, achieved gold-medal-level performance at the ICPC World Finals programming contest. |
| 25 | +- OpenAI reported a reasoning and code model achieved a **perfect score (12/12)** in ICPC testing. |
| 26 | +- Suno released **Suno v5**, an upgrade in music generation with studio-grade fidelity and more natural-sounding vocals. |
| 27 | +- Alibaba unveiled **Qwen-3-Max**, its flagship model with over a trillion parameters, focusing on long context and agent capabilities. |
| 28 | +- **Wan 2.2** was released, a generative video model focused on multi-shot consistency and character animation. |
| 29 | +- Anthropic announced **Claude Sonnet 4.5**, a model optimized for coding, agent construction, and improved reasoning. |
| 30 | +- OpenAI released **Sora 2**, a flagship video and audio generation model with improved physical modeling and synchronized sound. |
| 31 | +- DeepSeek released **DeepSeek-V3.2-Exp** |
| 32 | +- OpenAI and NVIDIA announced a strategic partnership for NVIDIA to supply at least **10 gigawatts** of AI systems for OpenAI's infrastructure. |
0 commit comments