The model maker just became a chip maker. openai's jalapeño inference chip, detailed oct 4, claims…
the model maker just became a chip maker. openai's jalapeño inference chip, detailed oct 4, claims 3.6x lower latency than nvidia silicon by tuning blocks to transformer inference.
the stack now bends all the way to the wafer.
Context
Verified event dates: OpenAI and Broadcom unveiled Jalapeño, OpenAI's first custom inference chip, on Jun 24, 2026. OpenAI published first results on Aug 25, 2026, after presenting at Hot Chips, claiming 1.5x to 1.9x more AI work per watt and 1.7x to 3.6x lower end-to-end latency than the best commercially available systems.
Tom's Hardware (Aug 25) reports the comparison was against Nvidia's GB200 and GB300 rack systems on SemiAnalysis's InferenceX suite, with a 700W Jalapeño against parts rated 1,200W and 1,400W, and notes the comparisons used single-token prediction while Nvidia deployments commonly use multi-token prediction.
The 3.6x lower latency is the top of OpenAI's 1.7x to 3.6x range, against Nvidia's GB200 and GB300 systems, and the chip was built with Broadcom. The post says 'detailed oct 4'; the pages read date the unveil Jun 24 and the first results Aug 25. 'Claims' is right: the figures are OpenAI's benchmarks, with caveats noted by Tom's Hardware. 'The stack now bends all the way to the wafer' is the author's.
'the model maker just became a chip maker' is the author's opinion.
Related work
Watch next
- Check whether SemiAnalysis published its own Jalapeño numbers.
Sources
- OpenAI, Jalapeño first results, Aug 25, 2026openai.com
- ServeTheHome, Jalapeño at Hot Chips 2026servethehome.com
- The Decoder, Jalapeño benchmarksthe-decoder.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 11 October 2026 at 10:44 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →