←
Samudro25 posts
Samudro avatar

Samudro

@PacificWNL

Engineering, Jarvislabs.ai

◷ Joined July 2020▧ 1 Photos and videos
Atharva Ingle avatar
⟳ Samudro retweeted
Atharva Ingle@AtharvaIngle7·22h

We launched Jarvislabs Token Factory two months ago, running on GPUs in India, and have been heads down improving it ever since. We benchmarked the same three open-weight models on Sarvam's Token Factory and ours. We're on par with or faster than Sarvam across speed and latency metrics, while offering lower pricing. - GLM 5.3 Flash is almost 2x faster. - Gemma 4 31B and DeepSeek V4.1 Flash are slightly faster, with significantly lower pricing for input, cached, and output tokens. And we're running on H200s, not B200s btw.

Post attachment

Today, we are launching Sarvam Token Factory for everyone. We host our own models, and the full range of best-in-class open-weight models, in India. We match leading inference providers on performance and reliability. For enterprises, this means more models to choose from, and an AI stack they can own without lock-ins.

0:10
Noctus avatar
⟳ Samudro retweeted
Noctus@noctus91·Nov 7

Following up on the pricing research, the first round of performance testing is now done. Tested JarvisLabs and Sarvam AI across four models: DeepSeek V4.1 Flash, GLM 5.3 Flash, Gemma 4 31B and GLM 5.3. A few things stood out: JarvisLabs came out ahead on both TPS and TPS per rupee across 3 of the 4 models. while Sarvam was faster on Gemma 4 31B. One more interesting result: Sarvam's 105B Conversations model recorded the highest TPS in the entire test at 655 tokens/sec. Sharing the results below.

Post attachmentPost attachmentPost attachment
Vishnu - Jarvislabs.ai avatar
⟳ Samudro retweeted
Vishnu - Jarvislabs.ai@vishnuvig·Nov 3

If you want to do something similar, you can use @jarvislabsai managed endpoints. If you are interested in hosting GLM or Deepseek kind of models DM me.

BREAKING: Revolut moved ~2,000 engineers onto self-hosted GLM for coding and pushed Claude Code and Codex to the exception list. Per David Abitbol, who leads AI ops there: 56 Nvidia B200s, a blended 30 cents per million tokens, roughly a tenth of what they paid Anthropic and OpenAI. Engineering lead time improved 70%. Here's the part that matters: the gains showed up when tokens stopped being rationed. Engineers quit saving AI for the hard tickets and pointed it at everything. Frontier models become the exception, open weights become the default. That's a pricing problem for every lab heading to IPO. h/t @OliverMolander Would you move your team off frontier APIs if the bill dropped 90%?

Post attachment
rahul avatar
⟳ Samudro retweeted
rahul@raulkrr·Oct 31

now that i have some credits, thanks to @jarvislabsai @AtharvaIngle7 trying my best to be at top 50 till 4th nov, currently down to #52 [overall] also do checkout jarvislabs → their jl cli is awesome - agents can spin up h200, run tests and pause it themselves, no idle gpu spends and token Factory: pay-per-token opes source models at the same place these guys are awesome, go try them 🤍

rahul@raulkrrOct 30

#33 now!! in muse open kernel challenge its getting really tough to maintain a spot now.

Post attachment
Demon1 avatar
⟳ Samudro retweeted
Demon1@Demon1___·3 May 2024

🌸Demon1 x Zekken Giveaway 🌸 ✦ 5x Mystbloom Bundle ✦ To enter: 1⃣Follow @Demon1___ + @zekkenVAL + @YukiAim ✦ 2⃣Like + RT this tweet ✧ 🏆 5 Winners ✦ 📅Ends May 12, 2024 ✧

Post attachment
Justice for mom avatar
⟳ Samudro retweeted
Justice for mom@Justicefor12267·3 Mar 2024

My Mom Rekha Nagdevi, Aged 59 went for a routine walk and Grocery shopping at 7pm and never came back, Gist of an incident: While crossing road on 18th Feb 2024 at 7.42pm, she met with a road accident, where in a biker driving MH48 CH 0766 BMW G310-Son of Senior Govt official was above speed of 100 km/h with no senses, fully drunk dashed Bike into Stomach of Mom causing immediate death, multiple fractures and wounds on the body, Today is the 14thDay, and yet we are waiting for action from police towards arrest of an individual involved in this.(FIR No-0031/2024) When Asked, Authority says its section 304-A, bailable offence so what is the point in arresting the culprit. Even None of media/newspapers showcased this Brutal incident, I Requested Police authorities for medical report (MLC as per medical laws) of person involved in this accident as he too was injured, to check alcohol consumption, but again they were clueless, Observing such actions, I am scared now Whether Police is assisting the Culprit even in Tampering evidences and Guiding him towards safe zone. While doing a reference check of Culprit, it also came to my knowledge that its his 3rd Accident where in human life has been endangered. I expect Indian Judiciary System to support me against this Habituary act of accident who took my Mom’s life. She was our everything!! Justice Delayed=Denied @PMOIndia @mieknathshinde @DGPMaharashtra @lokmat @HindustanTimes @timesofindia @palgharmitra @collectorpal @Voiceofpalghar @BJP4India @ShivSenaUBT_ @rashtrapatibhvn @TNNavbharat @aajtak @RajThackeray @MSJEGOI #justice #accident #rashdriving #india #palghar @Dev_Fadnavis @AjitPawarSpeaks #Tweet #Shameful #RoadAccident

Post attachment
Samudro avatar
Samudro@PacificWNL·4 Apr 2023

All the updates and reworks and Vanguard shit is useless unless you are able to penalise the throwers who come and ruin others' games and day as well.