Eric Vishria
@ericvishria
Followers
18K
Following
8K
Media
40
Statuses
3K
GP @Benchmark. Director: @ConfluentInc @CerebrasSystems @Contentful @Benchling @CommerceLayer @acuitymd @FireworksAI_HQ @quilterai @pomerium_io @DeepCogito
Joined June 2008
This is the definition of VC that we aspire to.
And I have a different definition of “venture capitalist” than the world has recently adopted. To me, it’s someone whose personal involvement and efforts can materially impact your chance of success. The work not the $$. Others are investors. There are few true VCs in the world.
13
13
174
💊 Medicines are humanity’s greatest invention, yet the system for making them is broken. It still takes $2B and 10 yrs to get a drug to market. It’s time to rebuild biotech. #AI with the right data infrastructure is how we get there. https://t.co/Hp5FqGQBXh
benchling.com
Making a new medicine is too slow, too costly, and too uncertain. Read Benchling CEO and Cofounder Sajith Wickramasekara’s perspective on how AI presents a once-in-a-generation opportunity to fix a...
0
9
15
Witnessing @btaylor, @claybavor, and the team take @SierraPlatform from @benchmark's founding Series A in May 2023 to a $10B valuation today is a true 6-sigma event. It's the perfect entrepreneurial storm: generational founders meeting humanity's AI inflection point with a
Today, we’re announcing that we’ve raised $350M additional capital at a valuation of $10B, led by Greenoaks, which is doubling down on Sierra.
2
78
279
Another overnight success 😂
Remarkable acceleration at Duffel: after 6 years of building the best APIs to sell travel across flights and hotels, Duffel is now doing $30M+ of volume per month and growing quickly. Revenue run-rate will grow 8 - 10x this year. Amazing persistence by an incredible team!
1
0
16
In thinking through future implications of AI, it is more important than ever to be able to hold multiple opposing ideas in your head at once. Because the probability distribution for any outcome has widened and flattened, dramatically in many cases.
11
14
173
Yes.
I think the risk that companies build their own systems of record - ERP, ITSM, CRM etc - is incredibly low Companies are not stupid. They have no competence here, the stakes are massively high, and regardless of how easy it is, they would still have maintain and optimize it,
2
1
27
Today I’m debuting Parable. Successful companies make the impossible, inevitable. Their stories live long enough to become parables that inspire generations.
199
53
1K
Today, OpenAI returns to open source! 🚀 Excited to be a launch partner for gpt-oss, OpenAI’s new open-weight models with variable reasoning. First providers are @FireworksAI_HQ, @GroqInc, and @CerebrasSystems, with more to come! Here’s what's launching today 👇
36
70
1K
Imagine cutting the grant to one of the greatest living mathematicians
23
197
1K
@CerebrasSystems 2000 tokens/s is production-ready speed been testing similar throughput with mario - the difference between 100 tokens/s and 2000 isn't just faster responses, it's fundamentally different user experience at those speeds you can do real-time code generation, live debugging,
0
3
20
A small update - we had more traffic than anticipated. However, the endpoints are now scalable on Together AI for all models, including the 671B MoE. Test out the model here: https://t.co/Od1NXYVBxU (A huge thanks to the folks at @togethercompute for making this happen so
together.ai
671B mixture-of-experts model matching Deepseek R1 performance, 60% shorter reasoning chains, approaching o3 and Claude 4 capabilities
Today, we are releasing 4 hybrid reasoning models of sizes 70B, 109B MoE, 405B, 671B MoE under open license. These are some of the strongest LLMs in the world, and serve as a proof of concept for a novel AI paradigm - iterative self-improvement (AI systems improving themselves).
3
14
82
Qwen3-Coder at ~2000 tokens/sec is now live in Windsurf! ⚡️ Fully hosted on US servers by @CerebrasSystems. Video is 1x speed.
122
108
1K
Cerebras has been demonstrating its ability to host large MoEs at very high speeds this week, launching Qwen3 235B 2507 and Qwen3 Coder 480B endpoints at >1,500 output tokens/s ➤ @CerebrasSystems now offers endpoints for both Qwen3 235B 2507 Reasoning & Non-reasoning. Both
8
26
302
Every exec hire is an over reaction to the last hire in that position. That’s the point a lot of times and okay. But often CEOs take for granted the strengths of the last person, and over value compensating for those weaknesses. So being conscious of it is valuable.
1
0
19
@ericvishria @benchmark @adityaag @southpkcommons This cost-performance ratio sets a new benchmark - $3.5M for models rivaling Claude 4 Opus shows what focused open research can achieve.
0
1
2
SOTA, Western open models ✅ Very large models ✅ New dimension to scale up intelligence✅ Built for under $3.5M 🤯 Been awesome to work with the team since @benchmark's investment last November. Special thanks to @adityaag @southpkcommons for introducing us!
Today, we are releasing 4 hybrid reasoning models of sizes 70B, 109B MoE, 405B, 671B MoE under open license. These are some of the strongest LLMs in the world, and serve as a proof of concept for a novel AI paradigm - iterative self-improvement (AI systems improving themselves).
1
8
28
Who is your rival? https://t.co/HLNNZaVEq4 Seen this so many times in business too. @dittycheria led-BladeLogic made us better at Opsware. Snowflake makes Databricks better, and vice versa. Tom's honest reflection of how he felt then vs in retrospect is refreshing.
tombrady.com
0
2
28