You can also fine-tune Wan on Replicate. Use images to fine-tune with Ostris's AI toolkit (just like flux): https://
replicate.com/ostris/wan-lor
a-trainer/train
… Or use videos with @zsakib_
's tuner: https://
replicate.com/zsxkib/wan-lor
a-trainer/train
… These outputs were from the Cyberpunk 2077 image-to-video lora trained with this.
CODE
-
Fine-tune Wan model with images and videos on Replicate
By
–
-
Google Releases Gemini 2.5 Pro Experimental with 1M Token Context
By
–
Google released Gemini 2.5 Pro Experimental, the first model in its Gemini 2.5 family
— Rowan Cheung (@rowancheung) 26 mars 2025
—#1 on the LMArena
—SOTA capabilities across benchmarks for coding, math, science, and more
—Visual reasoning
—1M token context window (2M coming soon!)pic.twitter.com/3G7MfXsfxkGoogle released Gemini 2.5 Pro Experimental, the first model in its Gemini 2.5 family —#1 on the LMArena
—SOTA capabilities across benchmarks for coding, math, science, and more
—Visual reasoning
—1M token context window (2M coming soon!) -
AI Leaders Discuss Latest Platforms and Technologies
By
–
Very excited about the upcoming (and recent) AI leaders on the pod:
– @kevinweil (CPO of OpenAI)
– @8enmann (co-founder of Anthropic)
– @mntruell (CEO of Cursor)
– @_mohansolo (CEO of Codeium/Windsurf)
– @rauchg (CEO of Vercel/v0)
– @EricSimons (CEO of Bolt)
– @antonosika (CEO of -
Supercharging Bolt with Cursor Rules Guide
By
–
link to the write up: https://
bensbites.com/blog/superchar
ging-bolt-with-cursor-rules
… shoutout: @boltdotnew @AnthropicAI -
Building a Chrome Extension for Bolt with Claude
By
–
1hr 37mins of vibe coding
— Ben Tossell (@bentossell) 25 mars 2025
how i built a chrome extension for bolt with Bolt & Claude (to add your own feature rules)
on friday afternoon, i didn't know what to build so i sorta hit record and let claude figure it out as i built it.
(i was/am working on a cursor rules style… https://t.co/UK7nI2sTOc pic.twitter.com/pOddI3luZ11hr 37mins of vibe coding how i built a chrome extension for bolt with Bolt & Claude (to add your own feature rules) on friday afternoon, i didn't know what to build so i sorta hit record and let claude figure it out as i built it. (i was/am working on a cursor rules style
-
1 billion open source methods with LoRAs
By
–
You have 1 billion methods to do that in open source with LoRAs. Gone are the days when OpenAI released features never seen before.
-
Langgraph Computer Use Agent with OpenAI Model
By
–
Langgraph computer use agent!
— Harrison Chase (@hwchase17) 25 mars 2025
Use OpenAI’s new computer use model with Langgraph to get a computer use agent in only a few lines of code https://t.co/OdTKNpDLolLanggraph computer use agent! Use OpenAI’s new computer use model with Langgraph to get a computer use agent in only a few lines of code
-
Gemini 2.5 Pro Experimental Achieves Top Performance Benchmarks
By
–
Introducing Gemini 2.5 Pro Experimental! 🎉
— Oriol Vinyals (@OriolVinyalsML) 25 mars 2025
Our newest Gemini model has stellar performance across math and science benchmarks. It’s an incredible model for coding and complex reasoning, and it’s #1 on the @lmarena_ai leaderboard by a drastic 40 ELO margin. Only a handful of… pic.twitter.com/soiAVSD00jIntroducing Gemini 2.5 Pro Experimental! Our newest Gemini model has stellar performance across math and science benchmarks. It’s an incredible model for coding and complex reasoning, and it’s #1 on the @lmarena_ai leaderboard by a drastic 40 ELO margin. Only a handful of
-
Claude 2.5 Pro Achieves New SOTA Across AI Benchmarks
By
–
2.5 Pro sets new SOTA capabilities across benchmarks, including: — 18.8% on Humanity's Last Exam — 63.8% on SWE-Bench Verified (agentic coding)
— GPQA Diamond and AIME 2025 (STEM)
— Long-context and visual reasoning -
Claude 2.5 Pro: 1M Token Context and Multimodal Capabilities
By
–
2.5 Pro also comes with a 1M token context window (2M coming soon!) — and is able to process and understand massive datasets from multiple sources (text, audio, images, and video.
— Rowan Cheung (@rowancheung) 25 mars 2025
Google focused on coding performance, and it shows in the demos.
2.5 Pro creating an impressive… pic.twitter.com/RxX3AoWtx72.5 Pro also comes with a 1M token context window (2M coming soon!) — and is able to process and understand massive datasets from multiple sources (text, audio, images, and video. Google focused on coding performance, and it shows in the demos. 2.5 Pro creating an impressive