Log inSign up
Fastino Labs
87 posts
Image
user avatar
Fastino Labs
@fastinoAI
Agentic fine-tuning and inference for small (and large) language models.
Palo Alto, CA
fastino.ai
Joined July 2024
94
Following
767
Followers
  • Pinned
    user avatar
    Fastino Labs
    @fastinoAI
    Apr 21
    Today, we are launching Pioneer: the world’s first agent for fine-tuning and inferencing SLMs and LLMs. With Pioneer, you can fine-tune and deploy models like Qwen, Gemma, and Llama and achieve state-of-the-art performance in minutes, with a single prompt. Models are
    Image
    00:00
    2.9M
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 23
    GLM 5.2 sits only 1% behind Opus 4.8 on FrontierSWE and 3% on PostTrainBench. A competitive open source model for long-horizon coding tasks has finally arrived. And it's now available for inference on Pioneer: pioneer.ai
    Image
    364
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 22
    “Tokenmaxxing” is no longer a sustainable way to measure productivity or success. Cost of tokens is rising, and even the simplest tasks burn hundreds to thousands of tokens running through the most expensive models when they don’t need to. Last week, our co-founder @ash_csx
    Image
    344
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 18
    Pioneer spotted in Germany 🔎 We spent last weekend in Munich at Tacto headquarters for @techeurope_'s hackathon, where we saw agents built for everything from procurement to on-device screen intelligence. Congratulations to the winner of our side challenge, and thank you to
    Image
    00:00
    214
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 17
    More than 100 of you joined our first lightning talk event, where conversations ranged from automating workflows to improving token efficiency. Thank you to the team at @ExaAILabs for co-hosting with us, and to our speakers @ash_csx, @DhruvAtreja1, and @fine_whines for their
    Image
    00:00
    195
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 15
    👋 See you there
    user avatar
    RAISE Summit
    @RaiseSummit
    Jun 15
    AI used to be a model you called. Now it's a stack you build: the database, the data center, the voice, the agent, and everything connecting them. We are proud to welcome these Official Partners to RAISE Summit 2026, building the infrastructure powering the next generation of
    Image
    208
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 15
    Last week, the Fastino Labs team headed up to San Francisco for the @tokensandai Harness Engineering Hackathon. @DhruvAtreja1, our Head of Agent Research, and Matthew Thomas, Technical Member of Staff, were on the ground demoing Pioneer and answering questions from teams
    Image
    00:00
    335
  • Fastino Labs reposted
    user avatar
    Thao N.
    @thaooowinnn
    Jun 12
    The line was out of the door at Harness Eng Hackathon organized by @tokensandai today 🤯 I’ll be here all day with @fastinoAI team - come say hi if you’re here 👋 psa: we have cute fox stickers 🦊
    Image
    00:00
    Image
    Image
    271
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 12
    We're heading up to SF on June 18th for the @tokensandai Agentic Engineering Summit. The agenda is stacked. We'll be there alongside companies like @AnthropicAI, @awscloud, @vercel, @stripe, @twilio , and @render. If you'll be attending, make sure to catch @ash_csx speaking
    Image
    228
  • Fastino Labs reposted
    user avatar
    Nikhil Nayak
    Fastino Labs
    @nikhilnayak268
    Jun 12
    1/8 We published a new paper from @fastinoAI: Correcting Stochastic Update Bias in Preconditioned Language Model Optimizers. Main idea: adaptive optimizers like AdamW, Sophia, and Shampoo do not just have noisy updates. Their stochastic preconditioned updates are biased in
    Image
    5.8K
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 11
    We’re less than one week away from our happy hour with @ExaAILabs on June 16th in SF. Hear from speakers on the importance of continual learning, and how that will impact the future of both model improvement and agentic search, followed by some networking over food and drinks.
    Image
    128
  • user avatar
    Fastino Labs
    @fastinoAI
    Jun 10
    Pioneer now supports Claude Fable 5. Anthropic’s newest model, Fable 5 shows advanced performance on long, complex tasks like software engineering, knowledge work, vision, and scientific research. On SWE-Bench Pro, it scores 80.3%, in comparison to 69.2% previously scored by
    Image
    330
  • Fastino Labs reposted
    user avatar
    tokens&
    @tokensandai
    Jun 9
    ever wondered how AI models become production-ready? this friday. come build with us master the tools, systems, and feedback loops that turn AI models into reliable agents. 💵 $40K+ in prizes (cash,credits, giftcards) 📅 June 12 · SF 🎟️ Teams of 4 max @awscloud
    Image
    516

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up