I test AI tools in real-world use and share what actually works for me. I also talk about technology and other topics that interest me. @Hitvox
Brasil
Joined August 2009
- Tweets1.7K
- Following4.6K
- Followers657
- Likes16.1K
DeepSeek has released DeepSeek-V4.1-Flash-1017, a minor update focused on reasoning, coding, agentic tasks and visual benchmarks. The model supports a 1M-token context window, reduces execution time and output length.
🚀 DeepSeek-V4.1-Flash-1017 is here!
Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.
🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.
1/4
Daivison Cardoso retweeted
After d1, we didn't want to chase incremental improvements.
We went back to the drawing board and redesigned it from scratch: data, model, evals, serving, all backed by ablations.
The result is d1-ultra, the world's best decision model! 🥳
Today we release 𝚍𝟷-𝚞𝚕𝚝𝚛𝚊, our most powerful decision model.
> built for hard decision problems.
> d1-ultra is multimodal
> supports 44 languages
> #1 on the @ArtificialAnlys Decision Index (65)
> leads all eight categories of our 112-evaluation suite
1/
Liquid AI has announced d1-ultra, its new multimodal decision model for complex real-world tasks. It supports text and images, ranks first on the Artificial Analysis Decision Index and is available through the Liquid API for $0.042 per million input tokens, with no output-token charges.
Today we release 𝚍𝟷-𝚞𝚕𝚝𝚛𝚊, our most powerful decision model.
> built for hard decision problems.
> d1-ultra is multimodal
> supports 44 languages
> #1 on the @ArtificialAnlys Decision Index (65)
> leads all eight categories of our 112-evaluation suite
1/
StepFun just open-sourced Step 5 Preview, a 600B-parameter sparse MoE model with 27B active parameters and up to 1M tokens of context. It supports text, images and video, with a focus on coding, long-horizon agents, professional work and finance.
Step 5 Preview is now open weights.
BF16 and FP8 weights are live on Hugging Face and ModelScope, together with our technical blog.
• Sparse MoE: 600B total, 27B active per token
• 1M-token context window
• Native text and vision input
• Built for long-horizon agentic work: coding, software engineering, professional knowledge work and finance
Frontier intelligence, open weights. Build what's next.
Hugging Face: huggingface.co/stepfun-ai/St…
ModelScope: modelscope.ai/models/stepfun…
GitHub: github.com/stepfun-ai/Step-5…
Daivison Cardoso retweeted
Try instabox.sh now!
We give your AI agents their own cloud computers.
Early access includes 20 vCPU, 40GB RAM for free.
Comment or DM me for invite code.
Stop using your laptop / Mac mini / VPS
Introducing instabox.sh, cloud computers for AI agents:
- 4 vCPU, 8 GB RAM spun up in <1s
- Built-in computer use
- Snapshots for easy cloning
- Persistent files, packages, config
- Public URLs for local ports
We're launching today through an invite-only beta program.
Early access includes 20 vCPU, 40GB RAM for free.
If you want to try, comment or DM me.
Daivison Cardoso retweeted
Introducing Expand Fetch, a new way to scrape and crawl the web.
Your agent has never seen the web. No wonder it gets basic questions wrong.
Traditional Markdown APIs lose critical details. So we trained two System one models to change that:
Iris, a vision model that sees pages like you do, preserving tables, icons, and layouts.
X-Ray, a JSON model that uncovers the hidden data behind every page.
Every fetch also comes with a Wayback Machine. Every line links back to its source, so you can verify exactly what your agent saw.
Three products. One API.
On our benchmark of 1,094 questions across 326 challenging web pages, agents using Expand got 74% right. The next best API got 55%.
Available today at $1 per 1,000 fetches.
Try it at expand.ai
Daivison Cardoso retweeted
AI Videos finally sound real.
Introducing Mirelo SFX 2.0.
It watches your video, cuts the music you never prompted, keeps the dialogue, and builds a full sound design synced to the action and consistent across any length clips.
You’ve made it look right. We’ll make it sound right.
Live today through our API, on @runwayml and @fal.
Other platforms coming soon. (Thread)
This duo is really good. Hermes Agent and TinyFish make web research and automation feel much more practical.
ICYMI: Hermes can search and read the web for $0.
hermes-agent.nousresearch.co…
TinyFish is now a plugin in the @NousResearch Hermes Agent.
We pointed ours at the most boring job in the building: an Excel sheet full of empty cells.
TinyFish searched and fetched information for all the rows and handed back clean results to Hermes Agent to fill in the spreadsheet.
Gemini 4 Argon is built for complex, multi-step work across coding, finance, legal, cybersecurity and science. Google says it ranks first on the Vals Index and launches at $2 per million input tokens, $10 per million output tokens, with cached input 95% cheaper.
Meta introduced Muse Wonder and Muse Wonder Full, its most capable models yet for personal and professional agents. The focus goes beyond benchmarks, with Active Memory, sustained task execution, coding and safer actions across private user context.
SubQ Snap claims a strong lead over Jev on enterprise decision tasks, scoring 61.7 versus 57.9 on Decision Index v0.2.1 and outperforming on 13 of 16 benchmarks. It combines 384K tokens of context with $0.05 per million input tokens and free output.
Introducing SubQ Snap, the first long-context decision model for real enterprise work.
Snap is the only decision model or LLM where compute per token stays constant as context grows.
• 384K tokens of multimodal context, 12x larger than Jev.
• 61.7 on Decision Index v0.2.1 vs Jev’s 57.9, outperforming on 13 of 16 enterprise benchmarks.
•Processes 100K tokens and returns a decision in under one second.
•$0.05/MM input tokens. Output is free.
Available today to anyone.
The first 10,000 sign-ups get 100M free input tokens.
We’ll likely hit capacity fast and will add users as we add compute. Compute offers welcome 🙂.
Latency will be variable as we tune the platform to this new model and category of users.
Model speeds are 80ms for minimal context, 250ms for 32k, 660ms for 100k, and 2.5 seconds for 384k.
Visit: subq.ai/
Daivison Cardoso retweeted
We’re open-sourcing Drex 1.5 today!
The smallest #1 decision-making model, built for local and cloud deployment with a 128k context window.👨💻
Compatible with Ollama, Unsloth, OpenCode or plug it into Claude, Codex, Cursor, Hermes, Claw and more.
Try it now: nace.ai/developers
Cloud users: Claim a $25 sign up bonus
Daivison Cardoso retweeted
We solved SQLite's single-writer limitation.
✅ Multiple concurrent writers.
✅ Multiple processes.
✅ Same SQLite.
✅ No modifications.
marcobambini.substack.com/p/…
Prime Intellect rewrote Prime Agent from TypeScript to Rust using a swarm of 2,000+ agents across 10,000+ sandboxes and more than 200B tokens. The result is around 13x faster input readiness and over 80% lower startup memory usage.
Over two weeks, Prime Agent orchestrated a swarm of over 2,000 agents to rewrite itself in Rust.
With 10,000+ sandboxes, 200B+ GLM-5.3 tokens, and 16,000 agent-to-agent messages, this was our largest test yet of Prime Agent’s multi-agent capabilities and our infrastructure powering these agent swarms.
Prime Agent now reaches usable input ~13× faster and uses 83% less startup memory than before.
TinyFish is now an official Hermes Agent plugin. Search and Fetch are free, giving agents access to live web content at no cost. When a task needs real interaction, Browser and Web Agent can be enabled with a one-time authorization.
Big one for the Hermes Agent community.
TinyFish is now a plugin in the @NousResearch Hermes Agent.
Your Hermes agent gets free Search and Fetch, so it can find and read the live web at $0.
When a task needs real clicks, TinyFish Browser and Web Agent are one approval away.
Microsoft-Decision-1 is a purpose-built model for fast structured decisions, not a general-purpose chatbot. It targets routing, classification, prioritization and workflow control, delivering low latency at just $0.042 per million input tokens, with free output.
Introducing Microsoft-Decision-1, our new model for fast decision-making.
It delivers top performance on structured decision tasks, outperforming both LLMs and other decision models in latency and quality.
We’re already testing it across Microsoft for everything from incident response and quality control to scientific discovery.
Daivison Cardoso retweeted
Bun é Anthropic. Deno é Cloudflare. Typescript já era Microsoft. Mundo JS se consolidando nas Big Techs.
We are excited to announce that Deno is joining Cloudflare!
deno.com/blog/cloudflare
Deno is joining Cloudflare, bringing its team and work on distributed JavaScript applications closer to Workers and Durable Objects. The long-term vision is to make scalable server development simpler, with compute, storage and communication built into the programming model.
We are excited to announce that Deno is joining Cloudflare!
deno.com/blog/cloudflare