正在加载视频...

视频加载失败

Today, we’re introducing the Sarvam MCP Server. It gives coding agents a standard way to discover and build with Sarvam. All you need to is a add a single configuration, and Sarvam becomes available directly in your coding agent. It already knows the right models and tools, so you...

77,266 次观看 • 3 个月前 •via X (Twitter)

34 条评论

J.D. Rathore 🇮🇳 的头像
J.D. Rathore 🇮🇳3 个月前

Indus takes lots of time to give answers it is sometimes irritating. Please make it fast and one more thing Create the aesthetics of Indus crazzy good and unique.

Aravind Venugopal 的头像
Aravind Venugopal3 个月前

Wow, didn't know it was unannounced. Was trying to make use of it yesterday 😅

Somesh Sampat 的头像
Somesh Sampat3 个月前

We need a strong coding model from Sarvam ...

✦ SK ✦ 的头像
✦ SK ✦3 个月前

Please stick to voice models. again saying you just are showing how incapable we are in the text models and insulting us to the global scale.

Vipul Gupta 的头像
Vipul Gupta3 个月前

Any chance the team would be working on a more conventional, slight resurgent way to communicate... Maybe a CLI?

ArchX 的头像
ArchX3 个月前

Hey I’m a big fan of your work guys. I always think a SarvamAI and then a few shortcomings come out to me//help-me-nudge To begin with: 1. The moving away from the just an India image and to something global 2. How do we achieve: as step 1 we work and open prs to openclaw and hermes to get the sarvam as a config in- a big win 3. I am not sure if we are on openrouter - but a good avenue 4. We should now start showing how the models compare on benchmarks- not just the big benchmarks like Deep SWE or tooling etc- we made Sarvam diff with India specific nuances- so why not become selfish for once guys— start a team setup a India specific voice benchmark — if not done already and lets put this out there 5. We will not always come out best- given where anthropic and claude are— but when u see Chinese models come out showing metrics they dont compare to gpt or claude- they compare to their n-1 version; they bring out their improvements- even if someone calls this narrow or tunnel 6. We open up sarvam to skills— mcp is a great start — but /building-effective- agents-sarvam is a bug win 7. We need more data - remember PuchAI what they did — they just started a whats app chat and teated this out- though they were just a wrapper- I’m not sure where I’m going with this but we need to get this out there 8. Take Sarvam to colleges- setup Educational Benefits/ I do see u guys do hackathon but I never see the artifacts that came out- maybe the X algoritm :( 9. I just want to take us sooooo aheadd and there is just these small things we can do; and maybe I can contribute to the bigger Sarvam- sampaaan vision @SarvamAI / @vivek_raghavan

Pier Code 的头像
Pier Code3 个月前

For coding tasks, sarvam models can be used within as well! 😄

pasandida(:)aura 的头像
pasandida(:)aura3 个月前

@SarvamAI good work but what is the usecase of your mcp model? Why would i need this and what pain point does it solve for me as a user? What would happen if this mcp did not exist tommorow. Any thoughts on this??

Nibas 924 的头像
Nibas 9243 个月前

Feedback and emails to Indus by Sarvam AI remain unanswered. IndiaAI Mission uses 98.68 million rupees of public funds, but offers no consumer support. Consumer Support?

Inflectiv AI ⧉ 的头像
Inflectiv AI ⧉3 个月前

🙂🙂

Makaroni 的头像
Makaroni3 个月前

plug in one config and suddenly your agent speaks 10 indian languages thats lowkey the fastest regional AI integration ive seen

安叫兽|Bird🕊️ 🔶 BNB 的头像
安叫兽|Bird🕊️ 🔶 BNB3 个月前

这种接入方式对编码代理还挺省事的

Prashant 的头像
Prashant3 个月前

👍👍

Chandan Ganwani 的头像
Chandan Ganwani3 个月前

Tried for days to translate this simple clip into English using Sarvam. No result. Same messages go around in a loop.

Krishna S Upadhaya 的头像
Krishna S Upadhaya3 个月前

How? Can i use it in Antigravity?

Shaunak Srivastava 的头像
Shaunak Srivastava3 个月前

This is superb!

Vishnu | EthIndia 的头像
Vishnu | EthIndia3 个月前

looks impressive kudoa to sarvam team!

Baadshah 的头像
Baadshah3 个月前

Can sarvam provide more fast answer

Om Keswani 的头像
Om Keswani3 个月前

One config instead of custom integrations? That’s the kind of developer experience that drives adoption. 🚀

Kaveri Engine 的头像
Kaveri Engine3 个月前

@RajanAnandan 🔥🔥🔥🔥👏👏👏

Swasti Pahuja 的头像
Swasti Pahuja3 个月前

Lets try making cool stuff using this now onwards….

Aaditya R. Patro | Founder, Inference 的头像
Aaditya R. Patro | Founder, Inference3 个月前

Much needed one. Thanks @SarvamAI

CyreneAI 的头像
CyreneAI3 个月前

Sarvam Progressing well 👀

Rehan Momin 的头像
Rehan Momin3 个月前

Let's make @SarvamAI launches more cool! And the community side more better with more events

Narendra 的头像
Narendra3 个月前

Great Move!🎉💯

Brosko 的头像
Brosko3 个月前

ngl the 'create key, copy config, build' pipeline is genuinely the right amount of steps anything more and devs start ghosting the docs 💀

Dimple 的头像
Dimple3 个月前

One doubt…..Since heavy tasks like long-form audio translation..etc so it runs asynchro…. ???

Photomk Phot 的头像
Photomk Phot3 个月前

instead a cli and agent skills would saved token , when mcp is getting it down you build it. Vaazhga janannayagam

Max 🚀 的头像
Max 🚀3 个月前

Damn this is impressive to be honest.

Shivam Rana 的头像
Shivam Rana2 个月前

Congrats! Some time back you guys mentioned that a coding specific model will also be launched. I am quite excited for that. Hopefully something is shared soon?🤞

Manav Bhatia 的头像
Manav Bhatia3 个月前

I am definitely going to use it via MCP for development

Vikram Singh 的头像
Vikram Singh3 个月前

@a_kraken_head

Rohan Patil 的头像
Rohan Patil3 个月前

Love seeing more companies embrace MCP. A standard integration layer saves developers time and lets them focus on building instead of wiring APIs together.

Sajan Singh 的头像
Sajan Singh2 个月前

Grow fast 🫀indian intelligence

相关视频