Merge pull request #36 from NHLOCAL/update

Update August

Author: NHLOCAL

Gemfile

@@ -1,2 +1,5 @@
 source "https://rubygems.org"
-gem "jekyll"
+gem "jekyll"
+gem "csv"
+gem "logger"
+gem "base64"

Gemfile.lock

@@ -4,9 +4,9 @@ GEM
     addressable (2.8.7)
       public_suffix (>= 2.0.2, < 7.0)
     base64 (0.3.0)
-    bigdecimal (3.2.2)
+    bigdecimal (3.1.8)
     colorator (1.1.0)
-    concurrent-ruby (1.3.5)
+    concurrent-ruby (1.3.4)
     csv (3.3.5)
     em-websocket (0.5.3)
       eventmachine (>= 0.12.9)
@@ -56,6 +56,7 @@ GEM
     listen (3.9.0)
       rb-fsevent (~> 0.10, >= 0.10.3)
       rb-inotify (~> 0.9, >= 0.9.10)
+    logger (1.7.0)
     mercenary (0.4.0)
     pathutil (0.16.2)
       forwardable-extended (~> 2.6)
@@ -81,7 +82,10 @@ PLATFORMS
   x86_64-linux
 
 DEPENDENCIES
+  base64
+  csv
   jekyll
+  logger
 
 BUNDLED WITH
    2.6.9

_data/timeline.md

@@ -59,7 +59,7 @@
 - Adobe releases **Firefly 2**.
 
 ## November
-- **Stable Diffusion XL Turbo** is released - A fast model that allows the creation of an image in one step in real time.
+- **Stable Diffusion XL Turbo** is released - A fast model that allows the creation of an image in one step in real-time.
 
 ## December
 - **Midjourney v6** is launched.
@@ -244,7 +244,7 @@
 ## June 2025
 - Google releases **Gemini 2.5 Pro** (final production-ready version), which leads benchmarks across the board.
 - ElevenLabs rolls out **Eleven v3 (alpha)** TTS with fine grained emotion control and support for 70+ languages.
-- OpenAI debuts **o3 pro**, an enhanced reasoning model offering extended context and real time tool integrations.
+- OpenAI debuts **o3 pro**, an enhanced reasoning model offering extended context and real-time tool integrations.
 
 ## July 2025
 - xAI releases **Grok 4**, achieving a new SOTA of 15.9% on ARC-AGI v2 and 25.4% on Humanity’s Last Exam. (*special*)
@@ -253,4 +253,13 @@
 - Google introduces **Gemini Deep Think**, which also earns an IMO 2025 gold by solving five of six problems with parallel reasoning. (*special*)
 - Alibaba open-sources two variants, **Qwen3-235B-A22B-Instruct-2507** (instruction-tuned) and **Qwen3-Coder**, for general LLM use and automated code generation.
 - Moonshot AI debuts **Kimi K2**, a Chinese LLM praised for its open-research focus and robust performance.
-- Chinese startup Zhipu open-sources **GLM-4.5**, a 130 B-parameter model tailored for intelligent-agent applications.
+- Chinese startup Zhipu open-sources **GLM-4.5**, a 130 B-parameter model tailored for intelligent-agent applications.
+
+## August 2025
+- Google introduced **Gemini 2.5 Deep Think**, a special "extended thinking" mode for solving complex problems and exploring alternatives. (*special*)
+- Anthropic released **Claude Opus 4.1**, an upgrade focused on improving agentic capabilities and real-world coding.
+- Google DeepMind announced **Genie 3.0**, a "world model" for creating interactive 3D environments from text, maintaining consistency for several minutes. (*special*)
+- 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.
+- OpenAI launched **GPT-5**, the company's next-generation model, with significant improvements in coding and a dynamic "thinking" mode to reduce hallucinations.
+- DeepSeek released **DeepSeek V3.1**, a hybrid model combining fast and slow "thinking" modes to improve performance in agentic tasks and tool use.
+- 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*)

_data/timeline.yml

@@ -62,7 +62,7 @@
     - text: Adobe releases <b>Firefly 2</b>.
   - date: November
     info:
-    - text: <b>Stable Diffusion XL Turbo</b> is released - A fast model that allows the creation of an image in one step in real time.
+    - text: <b>Stable Diffusion XL Turbo</b> is released - A fast model that allows the creation of an image in one step in real-time.
   - date: December
     info:
     - text: <b>Midjourney v6</b> is launched.
@@ -278,7 +278,7 @@
     info:
     - text: Google releases <b>Gemini 2.5 Pro</b> (final production-ready version), which leads benchmarks across the board.
     - text: ElevenLabs rolls out <b>Eleven v3 (alpha)</b> TTS with fine grained emotion control and support for 70+ languages.
-    - text: OpenAI debuts <b>o3 pro</b>, an enhanced reasoning model offering extended context and real time tool integrations.
+    - text: OpenAI debuts <b>o3 pro</b>, an enhanced reasoning model offering extended context and real-time tool integrations.
   - date: July 2025
     info:
     - text: xAI releases <b>Grok 4</b>, achieving a new SOTA of 15.9% on ARC-AGI v2 and 25.4% on Humanity’s Last Exam.
@@ -292,3 +292,15 @@
     - text: Alibaba open-sources two variants, <b>Qwen3-235B-A22B-Instruct-2507</b> (instruction-tuned) and <b>Qwen3-Coder</b>, for general LLM use and automated code generation.
     - text: Moonshot AI debuts <b>Kimi K2</b>, a Chinese LLM praised for its open-research focus and robust performance.
     - text: Chinese startup Zhipu open-sources <b>GLM-4.5</b>, a 130 B-parameter model tailored for intelligent-agent applications.
+  - date: August 2025
+    info:
+    - text: Google introduced <b>Gemini 2.5 Deep Think</b>, a special "extended thinking" mode for solving complex problems and exploring alternatives.
+      special: true
+    - text: Anthropic released <b>Claude Opus 4.1</b>, an upgrade focused on improving agentic capabilities and real-world coding.
+    - text: Google DeepMind announced <b>Genie 3.0</b>, a "world model" for creating interactive 3D environments from text, maintaining consistency for several minutes.
+      special: true
+    - text: OpenAI released <b>gpt-oss-120b</b> and <b>gpt-oss-20b</b>, a family of open-source models with high reasoning capabilities, optimized to run on accessible hardware.
+    - text: OpenAI launched <b>GPT-5</b>, the company's next-generation model, with significant improvements in coding and a dynamic "thinking" mode to reduce hallucinations.
+    - text: DeepSeek released <b>DeepSeek V3.1</b>, a hybrid model combining fast and slow "thinking" modes to improve performance in agentic tasks and tool use.
+    - text: Google launched a preview of <b>Gemini 2.5 Flash Image</b> (showcased as *nano-banana*), an advanced model for precise image editing, merging, and maintaining character consistency.
+      special: true

notes.md

@@ -23,5 +23,11 @@
 * Chinese startup Zhipu open-sources **GLM-4.5**, a 130 B-parameter model tailored for intelligent-agent applications.
 
 
-## אוגוסט 2025
-- גוגל שחררה את ג'מיני 2.5 DEEP THINK, המשלב חשיבה ממושכת ומציג ביצועי SOTA במגוון מדדים
+## August 2025
+- Google introduced **Gemini 2.5 Deep Think**, a special "extended thinking" mode for solving complex problems and exploring alternatives. (*special*)
+- Anthropic released **Claude Opus 4.1**, an upgrade focused on improving agentic capabilities and real-world coding.
+- Google DeepMind announced **Genie 3.0**, a "world model" for creating interactive 3D environments from text, maintaining consistency for several minutes. (*special*)
+- 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.
+- OpenAI launched **GPT-5**, the company's next-generation model, with significant improvements in coding and a dynamic "thinking" mode to reduce hallucinations. (*special*)
+- DeepSeek released **DeepSeek V3.1**, a hybrid model combining fast and slow "thinking" modes to improve performance in agentic tasks and tool use.
+- 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*)

start.bat

@@ -1 +1,3 @@
+python scripts/convert_timeline_events.py _data/timeline.md
+
 bundle exec jekyll serve