Ai News and Reviews
@softreviewed.bsky.social
Follow me for the latest AI updates and news
▶️ https://www.youtube.com/@softreviewed
🔗 https://softreviewed.com/
▶️ https://www.youtube.com/@softreviewed
🔗 https://softreviewed.com/
AI writes essays, but software workflows need instant reflexes.
TypeSafe AI just launched Jev: a dedicated System 1 decision model delivering 70ms responses, 4¢/1M input tokens, and free outputs. Here is how it compares to frontier LLMs:
TypeSafe AI just launched Jev: a dedicated System 1 decision model delivering 70ms responses, 4¢/1M input tokens, and free outputs. Here is how it compares to frontier LLMs:
September 26, 2026 at 5:06 PM
AI writes essays, but software workflows need instant reflexes.
TypeSafe AI just launched Jev: a dedicated System 1 decision model delivering 70ms responses, 4¢/1M input tokens, and free outputs. Here is how it compares to frontier LLMs:
TypeSafe AI just launched Jev: a dedicated System 1 decision model delivering 70ms responses, 4¢/1M input tokens, and free outputs. Here is how it compares to frontier LLMs:
Run an entire online business autonomously with Systeme.io & Antigravity! 🚀⚡
130 native e-commerce & marketing endpoints controlled via conversational AI agents. Here is how it works:
130 native e-commerce & marketing endpoints controlled via conversational AI agents. Here is how it works:
September 25, 2026 at 8:15 PM
Run an entire online business autonomously with Systeme.io & Antigravity! 🚀⚡
130 native e-commerce & marketing endpoints controlled via conversational AI agents. Here is how it works:
130 native e-commerce & marketing endpoints controlled via conversational AI agents. Here is how it works:
Google AI Pro + 5 TB storage shared across 5 family accounts for $0.
Individual private storage, zero shared chat history, and Gemini 3 Pro reasoning. Here is how family pooling works:
Individual private storage, zero shared chat history, and Gemini 3 Pro reasoning. Here is how family pooling works:
September 24, 2026 at 10:57 AM
Google AI Pro + 5 TB storage shared across 5 family accounts for $0.
Individual private storage, zero shared chat history, and Gemini 3 Pro reasoning. Here is how family pooling works:
Individual private storage, zero shared chat history, and Gemini 3 Pro reasoning. Here is how family pooling works:
Share Google AI Pro with 5 family members! 🚀⚡
• 6 total people share 1 subscription
• 5 TB shared storage pool
• Gemini 3 Pro + Workspace AI for all
• Antigravity quotas refreshed
• Jio 5G 18-month free offer explained
Full tutorial in reply 👇
• 6 total people share 1 subscription
• 5 TB shared storage pool
• Gemini 3 Pro + Workspace AI for all
• Antigravity quotas refreshed
• Jio 5G 18-month free offer explained
Full tutorial in reply 👇
September 23, 2026 at 11:44 AM
Share Google AI Pro with 5 family members! 🚀⚡
• 6 total people share 1 subscription
• 5 TB shared storage pool
• Gemini 3 Pro + Workspace AI for all
• Antigravity quotas refreshed
• Jio 5G 18-month free offer explained
Full tutorial in reply 👇
• 6 total people share 1 subscription
• 5 TB shared storage pool
• Gemini 3 Pro + Workspace AI for all
• Antigravity quotas refreshed
• Jio 5G 18-month free offer explained
Full tutorial in reply 👇
OpenAI officially released GPT-6 Sol and GPT-6 Luna! 🚀⚡
• GPT-6 Sol: 50% price reduction ($2/$10 per 1M tokens)
• 90% prompt caching cut down to $0.20/M
• 1.05M context window (922K in / 128K out)
• 33.2% AutomationBench autonomous coding
• GPT-6 Luna: $0.10/M utility tier at 175 tok/s
Full archi…
• GPT-6 Sol: 50% price reduction ($2/$10 per 1M tokens)
• 90% prompt caching cut down to $0.20/M
• 1.05M context window (922K in / 128K out)
• 33.2% AutomationBench autonomous coding
• GPT-6 Luna: $0.10/M utility tier at 175 tok/s
Full archi…
September 23, 2026 at 10:56 AM
OpenAI officially released GPT-6 Sol and GPT-6 Luna! 🚀⚡
• GPT-6 Sol: 50% price reduction ($2/$10 per 1M tokens)
• 90% prompt caching cut down to $0.20/M
• 1.05M context window (922K in / 128K out)
• 33.2% AutomationBench autonomous coding
• GPT-6 Luna: $0.10/M utility tier at 175 tok/s
Full archi…
• GPT-6 Sol: 50% price reduction ($2/$10 per 1M tokens)
• 90% prompt caching cut down to $0.20/M
• 1.05M context window (922K in / 128K out)
• 33.2% AutomationBench autonomous coding
• GPT-6 Luna: $0.10/M utility tier at 175 tok/s
Full archi…
Claude Opus 5.5 released by Anthropic! 🚀⚡
• 66.4% Terminal-Bench record (#1 in coding)
• Fable 5.1 power at 40% lower cost ($4/$20)
• 1,846 Elo on GDPval-AA (+304 over GPT-6 Astra)
• $0.20/M cache reads (-60% price drop)
• +20% 5-hour limit boost in Claude Code
Full architectural review in reply 👇
• 66.4% Terminal-Bench record (#1 in coding)
• Fable 5.1 power at 40% lower cost ($4/$20)
• 1,846 Elo on GDPval-AA (+304 over GPT-6 Astra)
• $0.20/M cache reads (-60% price drop)
• +20% 5-hour limit boost in Claude Code
Full architectural review in reply 👇
September 22, 2026 at 6:56 PM
Claude Opus 5.5 released by Anthropic! 🚀⚡
• 66.4% Terminal-Bench record (#1 in coding)
• Fable 5.1 power at 40% lower cost ($4/$20)
• 1,846 Elo on GDPval-AA (+304 over GPT-6 Astra)
• $0.20/M cache reads (-60% price drop)
• +20% 5-hour limit boost in Claude Code
Full architectural review in reply 👇
• 66.4% Terminal-Bench record (#1 in coding)
• Fable 5.1 power at 40% lower cost ($4/$20)
• 1,846 Elo on GDPval-AA (+304 over GPT-6 Astra)
• $0.20/M cache reads (-60% price drop)
• +20% 5-hour limit boost in Claude Code
Full architectural review in reply 👇
Grok 4.7 released by xAI! 🚀
• 2.1T MoE parameter scale
• 500k context window
• The RL quitting flaw explained
• 64.0% EEBench (#1 over Claude Fable & GPT-5.6)
• 71.0% DeepSWE
• $2/$6 per 1M tokens
Full guide in reply 👇
• 2.1T MoE parameter scale
• 500k context window
• The RL quitting flaw explained
• 64.0% EEBench (#1 over Claude Fable & GPT-5.6)
• 71.0% DeepSWE
• $2/$6 per 1M tokens
Full guide in reply 👇
September 21, 2026 at 7:35 PM
Grok 4.7 released by xAI! 🚀
• 2.1T MoE parameter scale
• 500k context window
• The RL quitting flaw explained
• 64.0% EEBench (#1 over Claude Fable & GPT-5.6)
• 71.0% DeepSWE
• $2/$6 per 1M tokens
Full guide in reply 👇
• 2.1T MoE parameter scale
• 500k context window
• The RL quitting flaw explained
• 64.0% EEBench (#1 over Claude Fable & GPT-5.6)
• 71.0% DeepSWE
• $2/$6 per 1M tokens
Full guide in reply 👇
Pictory MCP for Free: Automate AI video generation directly inside your IDE. Connect Claude Desktop and Cursor to Pictory for instant script-to-video creation and text-based editing.
September 15, 2026 at 7:08 PM
Pictory MCP for Free: Automate AI video generation directly inside your IDE. Connect Claude Desktop and Cursor to Pictory for instant script-to-video creation and text-based editing.
Google just launched Gemini for Windows as a standalone desktop client.
• Summon instantly anywhere via Alt + Space
• 1-click active screen capture & multimodal analysis
• Drag-and-drop local files & code
• Powered by Gemini 2.5 with 2M context
Full technical breakdown in reply 👇
• Summon instantly anywhere via Alt + Space
• 1-click active screen capture & multimodal analysis
• Drag-and-drop local files & code
• Powered by Gemini 2.5 with 2M context
Full technical breakdown in reply 👇
September 12, 2026 at 9:47 AM
Google just launched Gemini for Windows as a standalone desktop client.
• Summon instantly anywhere via Alt + Space
• 1-click active screen capture & multimodal analysis
• Drag-and-drop local files & code
• Powered by Gemini 2.5 with 2M context
Full technical breakdown in reply 👇
• Summon instantly anywhere via Alt + Space
• 1-click active screen capture & multimodal analysis
• Drag-and-drop local files & code
• Powered by Gemini 2.5 with 2M context
Full technical breakdown in reply 👇
Meta just launched Muse—a 24/7 personal cloud agent with 100M free tokens/week, confidential VMs co-designed with Moxie Marlinspike, and native Stripe/Shopify execution.
👇 Full architectural breakdown and review in the first reply!
👇 Full architectural breakdown and review in the first reply!
September 9, 2026 at 6:26 PM
Meta just launched Muse—a 24/7 personal cloud agent with 100M free tokens/week, confidential VMs co-designed with Moxie Marlinspike, and native Stripe/Shopify execution.
👇 Full architectural breakdown and review in the first reply!
👇 Full architectural breakdown and review in the first reply!
Meta launched Muse: a 24/7 personal cloud agent with 100M free tokens/wk, confidential VMs co-designed with Moxie Marlinspike, and native Stripe execution.
Review: https://softreviewed.com/meta-muse-ai-agent-review/
Watch: https://youtu.be/Uq5UbLf1a4w
Review: https://softreviewed.com/meta-muse-ai-agent-review/
Watch: https://youtu.be/Uq5UbLf1a4w
September 9, 2026 at 6:07 PM
Meta launched Muse: a 24/7 personal cloud agent with 100M free tokens/wk, confidential VMs co-designed with Moxie Marlinspike, and native Stripe execution.
Review: https://softreviewed.com/meta-muse-ai-agent-review/
Watch: https://youtu.be/Uq5UbLf1a4w
Review: https://softreviewed.com/meta-muse-ai-agent-review/
Watch: https://youtu.be/Uq5UbLf1a4w
OpenAI just launched ChatGPT Images 2.5.
• Surgical comment edits with zero face warping
• 50% lower generation latency
• Two new developer models (Flare & Sunburst)
• Native transparent PNG cutouts
Full technical guide in reply 👇
• Surgical comment edits with zero face warping
• 50% lower generation latency
• Two new developer models (Flare & Sunburst)
• Native transparent PNG cutouts
Full technical guide in reply 👇
September 8, 2026 at 8:07 PM
OpenAI just launched ChatGPT Images 2.5.
• Surgical comment edits with zero face warping
• 50% lower generation latency
• Two new developer models (Flare & Sunburst)
• Native transparent PNG cutouts
Full technical guide in reply 👇
• Surgical comment edits with zero face warping
• 50% lower generation latency
• Two new developer models (Flare & Sunburst)
• Native transparent PNG cutouts
Full technical guide in reply 👇
OpenAI launched GPT-6 Astra! You can get ChatGPT Plus completely free for a month.
• Eligibility check & activation
• Desktop app Astra setup
• Safe cancellation walkthrough
Full tutorial in reply 👇
• Eligibility check & activation
• Desktop app Astra setup
• Safe cancellation walkthrough
Full tutorial in reply 👇
September 8, 2026 at 2:20 PM
OpenAI launched GPT-6 Astra! You can get ChatGPT Plus completely free for a month.
• Eligibility check & activation
• Desktop app Astra setup
• Safe cancellation walkthrough
Full tutorial in reply 👇
• Eligibility check & activation
• Desktop app Astra setup
• Safe cancellation walkthrough
Full tutorial in reply 👇
ChatGPT is rolling out paid conversational search ads.
• Ads matched to real-time conversational context
• High intent solution-seeking placements
• New attribution & landing page dynamics
Full guide in reply 👇
• Ads matched to real-time conversational context
• High intent solution-seeking placements
• New attribution & landing page dynamics
Full guide in reply 👇
September 7, 2026 at 2:54 PM
ChatGPT is rolling out paid conversational search ads.
• Ads matched to real-time conversational context
• High intent solution-seeking placements
• New attribution & landing page dynamics
Full guide in reply 👇
• Ads matched to real-time conversational context
• High intent solution-seeking placements
• New attribution & landing page dynamics
Full guide in reply 👇
OpenAI made desktop voice dictation 100% FREE in the ChatGPT desktop app!
• System-wide speech to text (Word, Notion, browser)
• Hold-to-talk & hands-free toggle modes
• Custom dictionary fixes jargon typos
Full guide in reply 👇
• System-wide speech to text (Word, Notion, browser)
• Hold-to-talk & hands-free toggle modes
• Custom dictionary fixes jargon typos
Full guide in reply 👇
September 7, 2026 at 7:16 AM
OpenAI made desktop voice dictation 100% FREE in the ChatGPT desktop app!
• System-wide speech to text (Word, Notion, browser)
• Hold-to-talk & hands-free toggle modes
• Custom dictionary fixes jargon typos
Full guide in reply 👇
• System-wide speech to text (Word, Notion, browser)
• Hold-to-talk & hands-free toggle modes
• Custom dictionary fixes jargon typos
Full guide in reply 👇
OpenAI has officially launched GPT-6 Astra, its flagship foundation model pre-trained on 100,000+ GPUs at Stargate.
• 95.9% BenchCAD (#1 3D vision-to-CAD synthesis)
• 64.6% Terminal-Bench Science
• 99.9% ARC-AGI-3 visual inductive logic
• Autonomous computer-use in Blender, Excel & KiCad
• 100% of…
• 95.9% BenchCAD (#1 3D vision-to-CAD synthesis)
• 64.6% Terminal-Bench Science
• 99.9% ARC-AGI-3 visual inductive logic
• Autonomous computer-use in Blender, Excel & KiCad
• 100% of…
September 3, 2026 at 8:49 PM
OpenAI has officially launched GPT-6 Astra, its flagship foundation model pre-trained on 100,000+ GPUs at Stargate.
• 95.9% BenchCAD (#1 3D vision-to-CAD synthesis)
• 64.6% Terminal-Bench Science
• 99.9% ARC-AGI-3 visual inductive logic
• Autonomous computer-use in Blender, Excel & KiCad
• 100% of…
• 95.9% BenchCAD (#1 3D vision-to-CAD synthesis)
• 64.6% Terminal-Bench Science
• 99.9% ARC-AGI-3 visual inductive logic
• Autonomous computer-use in Blender, Excel & KiCad
• 100% of…
Meta just dropped Muse Spark 1.3, quietly taking the World #1 spot in autonomous AI coding.
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
September 3, 2026 at 2:47 PM
Meta just dropped Muse Spark 1.3, quietly taking the World #1 spot in autonomous AI coding.
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
Meta just dropped Muse Spark 1.3, quietly taking the World #1 spot in autonomous AI coding.
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
September 3, 2026 at 6:38 AM
Meta just dropped Muse Spark 1.3, quietly taking the World #1 spot in autonomous AI coding.
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
• 75.4% DeepSWE v1.1 (#1 worldwide pass rate)
• 88.8% Terminal-Bench 2.1
• 59.4% SWEAtlas monorepo comprehension
• 1M Context Horizon (98.1% needle retention)
• Flat $5/mo developer plans & $0.10/1M API tok…
Did you know your FREE ChatGPT account has a built-in browser? 🤯 Work side-by-side without switching tabs and 10x your workflow speed. Full free tutorial linked in the first comment 👇
September 2, 2026 at 8:11 PM
Did you know your FREE ChatGPT account has a built-in browser? 🤯 Work side-by-side without switching tabs and 10x your workflow speed. Full free tutorial linked in the first comment 👇
Gemini 3.8 Flash is here! Google packed Claude Opus 5 coding intelligence into a Flash tier at $0.75/$3.75 per 1M tokens.
89.4% on Terminal-bench 2.1 (#1 in the world). 85% cost collapse.
Full benchmark review in the reply below! 👇
89.4% on Terminal-bench 2.1 (#1 in the world). 85% cost collapse.
Full benchmark review in the reply below! 👇
September 2, 2026 at 5:51 PM
Gemini 3.8 Flash is here! Google packed Claude Opus 5 coding intelligence into a Flash tier at $0.75/$3.75 per 1M tokens.
89.4% on Terminal-bench 2.1 (#1 in the world). 85% cost collapse.
Full benchmark review in the reply below! 👇
89.4% on Terminal-bench 2.1 (#1 in the world). 85% cost collapse.
Full benchmark review in the reply below! 👇
Gemini 3.8 Flash is already running on select Google endpoints. The UI still says 3.7, but backend identity probes return 3.8 Flash directly. Free to test in Google AI Studio. Full verification method in the video.
September 2, 2026 at 7:16 AM
Gemini 3.8 Flash is already running on select Google endpoints. The UI still says 3.7, but backend identity probes return 3.8 Flash directly. Free to test in Google AI Studio. Full verification method in the video.
Anthropic officially released Claude Fable 5.1 and Claude Mythos 5.1!
• 52.6% Terminal-Bench-Science (2x Fable 5)
• 55.8% Terminal-Bench 4.0 Coding
• 75% cheaper cache reads ($0.25/M tokens)
• 1M context window + 128K max output
Full breakdown in reply 👇
• 52.6% Terminal-Bench-Science (2x Fable 5)
• 55.8% Terminal-Bench 4.0 Coding
• 75% cheaper cache reads ($0.25/M tokens)
• 1M context window + 128K max output
Full breakdown in reply 👇
September 1, 2026 at 8:34 PM
Anthropic officially released Claude Fable 5.1 and Claude Mythos 5.1!
• 52.6% Terminal-Bench-Science (2x Fable 5)
• 55.8% Terminal-Bench 4.0 Coding
• 75% cheaper cache reads ($0.25/M tokens)
• 1M context window + 128K max output
Full breakdown in reply 👇
• 52.6% Terminal-Bench-Science (2x Fable 5)
• 55.8% Terminal-Bench 4.0 Coding
• 75% cheaper cache reads ($0.25/M tokens)
• 1M context window + 128K max output
Full breakdown in reply 👇
Want to use 300+ AI models including free endpoints in your automations? Here is how to connect an OpenRouter API key to Make.com for text, image, and audio workflows without paying monthly subscription markups.
September 1, 2026 at 7:25 AM
Want to use 300+ AI models including free endpoints in your automations? Here is how to connect an OpenRouter API key to Make.com for text, image, and audio workflows without paying monthly subscription markups.
Wait... you can actually run GPT-5.6 Sol on a free ChatGPT account?! 🤯
👇 Full YouTube tutorial linked in the reply below:
👇 Full YouTube tutorial linked in the reply below:
August 30, 2026 at 1:19 PM
Wait... you can actually run GPT-5.6 Sol on a free ChatGPT account?! 🤯
👇 Full YouTube tutorial linked in the reply below:
👇 Full YouTube tutorial linked in the reply below:
Looking for a working Veed.io promo code? Get 50% OFF your first 3 months on Lite or Pro plans.
👇 Watch the full YouTube guide below!
👇 Watch the full YouTube guide below!
August 30, 2026 at 12:10 PM
Looking for a working Veed.io promo code? Get 50% OFF your first 3 months on Lite or Pro plans.
👇 Watch the full YouTube guide below!
👇 Watch the full YouTube guide below!