GPT‑5.3‑Codex‑Spark • Hacker News Board

Re: GPT‑5.3‑Codex‑Spark

#231

Post by tiffanyh » Fri, Feb 13, 2026, 1:09 AM UTC

pjs_ wrote: post #66

Continue to believe that Cerebras is one of the most underrated companies of our time. It's a dinner-plate sized chip. It actually works. It's actually much faster than anything else for real workloads. Amazing

If history has taught us anything, “engineered systems” (like mainframes & hyper converged infrastructure) emerge at the start of a new computing paradigm … but long-term, commodity compute wins the game.

Re: GPT‑5.3‑Codex‑Spark

#232

Post by i_love_retros » Fri, Feb 13, 2026, 1:13 AM UTC

perdomon wrote: post #183

This has been the industry standard for the last 20 minutes. I can't believe people are still using GPT-5.3-Codex.

Ha think of all the losers still using codex 5.2

Re: GPT‑5.3‑Codex‑Spark

#233

Post by elorant » Fri, Feb 13, 2026, 1:18 AM UTC

ElijahLynn wrote: post #210

Wow, I wish we could post pictures to HN. That chip is HUGE!!!! The WSE-3 is the largest AI chip ever built, measuring 46,255 mm² and containing 4 trillion transistors. It delivers 125 petaflops of AI compute through 900,000 AI-optimized cores — 19× more transistors and 28× more compute than the NVIDIA B200. From https://www.cerebras.ai/chip : https://cdn.sanity.io/images/e4qjo92p/production/78c94c67be9... https://cd…

There have been discussions about this chip here in the past. Maybe not that particular one but previous versions of it. The whole server if I remember correctly eats some 20KWs of power.

Re: GPT‑5.3‑Codex‑Spark

#234

Post by kumarvvr » Fri, Feb 13, 2026, 1:19 AM UTC

ElijahLynn wrote: post #210

Wow, I wish we could post pictures to HN. That chip is HUGE!!!! The WSE-3 is the largest AI chip ever built, measuring 46,255 mm² and containing 4 trillion transistors. It delivers 125 petaflops of AI compute through 900,000 AI-optimized cores — 19× more transistors and 28× more compute than the NVIDIA B200. From https://www.cerebras.ai/chip : https://cdn.sanity.io/images/e4qjo92p/production/78c94c67be9... https://cd…

Is this actually beneficial than, say having a bunch of smaller ones communicating on a bus? Apart from space constraints that is.

Re: GPT‑5.3‑Codex‑Spark

#235

Post by dateSISC » Fri, Feb 13, 2026, 1:26 AM UTC

Isn't this chip having 44GB sram total a big limitation for what it can run?

Re: GPT‑5.3‑Codex‑Spark

#236

Post by mpalmer » Fri, Feb 13, 2026, 1:36 AM UTC

Gemini Flash Lite 3 preview within 10 days now, surely

Re: GPT‑5.3‑Codex‑Spark

#237

Post by santaboom » Fri, Feb 13, 2026, 1:38 AM UTC

kumarvvr wrote: post #234

ElijahLynn wrote: post #210

Wow, I wish we could post pictures to HN. That chip is HUGE!!!! The WSE-3 is the largest AI chip ever built, measuring 46,255 mm² and containing 4 trillion transistors. It delivers 125 petaflops of AI compute through 900,000 AI-optimized cores — 19× more transistors and 28× more compute than the NVIDIA B200. From https://www.cerebras.ai/chip : https://cdn.sanity.io/images/e4qjo92p/production/78c94c67be9... https://cd…

Is this actually beneficial than, say having a bunch of smaller ones communicating on a bus? Apart from space constraints that is.

Yes, bandwidth within a chip is much higher than on a bus.

Re: GPT‑5.3‑Codex‑Spark

#238

Post by zamadatix » Fri, Feb 13, 2026, 1:38 AM UTC

kumarvvr wrote: post #234

ElijahLynn wrote: post #210

Wow, I wish we could post pictures to HN. That chip is HUGE!!!! The WSE-3 is the largest AI chip ever built, measuring 46,255 mm² and containing 4 trillion transistors. It delivers 125 petaflops of AI compute through 900,000 AI-optimized cores — 19× more transistors and 28× more compute than the NVIDIA B200. From https://www.cerebras.ai/chip : https://cdn.sanity.io/images/e4qjo92p/production/78c94c67be9... https://cd…

Is this actually beneficial than, say having a bunch of smaller ones communicating on a bus? Apart from space constraints that is.

It's a single wafer, not a single compute core. A familiar equivalent might be putting 192 cores in a single Epyc CPU (or, more to be more technically accurate, the group of cores in a single CCD) rather than trying to interconnect 192 separate single core CPUs externally with each other.

Re: GPT‑5.3‑Codex‑Spark

#239

Post by freakynit » Fri, Feb 13, 2026, 1:48 AM UTC

Does this prove cerebras chips are generic enough to be able to run the most common architectures of LLM's? Even the proprietary ones?

Re: GPT‑5.3‑Codex‑Spark

#240

Post by johnfn » Fri, Feb 13, 2026, 2:01 AM UTC

seunosewa wrote: post #120

johnfn wrote: post #76

Earlier quoted context omitted.

The other day I got Codex to one-shot an upgrade to Vite 8 at my day job (a real website with revenue). It worked in this for over 3 hours without intervention (I went to sleep). This is now in production.

How did you verify it?

Uhhh, this is my work, so… we didn’t have a SEV? None of our thousands of customers paying us money reported the site was broken?