As the designer and developer of this service, I'd love to get some feedback (and of course, early users). My company (Cerebris - I'm a co-founder) developed this application in order to try out a lot of tech, from Rails 3 to jQuery Mobile to Omniauth to hosting at Rackspace. Once we realized how useful it had become to us personally, we decided to add some extra polish and features.
We're keeping this service completely free, although we may add premium features in the future. We also plan to introduce native apps for a small fee (perhaps $1.99).
I really hope you find I.O.U. Mate fun and useful. Please let me know what you think - I'd appreciate feedback of any kind. Thanks in advance for checking it out!
http://IOUmate.com"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Never forget what you've loaned or borrowed with IOUmate.com"},"url":{"matchLevel":"none","matchedWords":[],"value":""}},"_tags":["story","author_dgeb","story_2532676","show_hn"],"author":"dgeb","children":[2532735],"created_at":"2011-05-10T14:55:59Z","created_at_i":1305039359,"num_comments":1,"objectID":"2532676","points":2,"story_id":2532676,"story_text":"I.O.U. Mate is a new service for tracking I.O.U.s of any kind, from dollars to drinks to books you've loaned. It can be used to schedule recurring debts like allowances or shared rent. You can invite friends to share I.O.U.s with Facebook, Twitter or email, or you can choose to track debts privately.
As the designer and developer of this service, I'd love to get some feedback (and of course, early users). My company (Cerebris - I'm a co-founder) developed this application in order to try out a lot of tech, from Rails 3 to jQuery Mobile to Omniauth to hosting at Rackspace. Once we realized how useful it had become to us personally, we decided to add some extra polish and features.
We're keeping this service completely free, although we may add premium features in the future. We also plan to introduce native apps for a small fee (perhaps $1.99).
I really hope you find I.O.U. Mate fun and useful. Please let me know what you think - I'd appreciate feedback of any kind. Thanks in advance for checking it out!
http://IOUmate.com","title":"Show HN: Never forget what you've loaned or borrowed with IOUmate.com","updated_at":"2024-09-19T17:42:07Z","url":""},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"I.O.U. Mate is a new service for tracking I.O.U.s of any kind, from dollars to drinks to books you've loaned. It can be used to schedule recurring debts like allowances or shared rent. You can invite friends to share I.O.U.s with Facebook, Twitter or email, or you can choose to track debts privately.
As the designer and developer of this service, I'd love to get some feedback (and of course, early users). My company (Cerebris - I'm a co-founder) developed this application in order to try out a lot of tech, from Rails 3 to jQuery Mobile to Omniauth to hosting at Rackspace. Once we realized how useful it had become to us personally, we decided to add some extra polish and features.
We're keeping this service completely free, although we may add premium features in the future. We also plan to introduce native apps for a small fee (perhaps $1.99).
I really hope you find I.O.U. Mate fun and useful. Please let me know what you think - I'd appreciate feedback of any kind. I'm particularly interested to know how people like the new mobile interface.
Thanks in advance for checking it out!
http://IOUmate.com"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Please review my site IOUmate.com"},"url":{"matchLevel":"none","matchedWords":[],"value":""}},"_tags":["story","author_dgeb","story_2533991","ask_hn"],"author":"dgeb","created_at":"2011-05-10T20:36:43Z","created_at_i":1305059803,"num_comments":0,"objectID":"2533991","points":2,"story_id":2533991,"story_text":"I.O.U. Mate is a new service for tracking I.O.U.s of any kind, from dollars to drinks to books you've loaned. It can be used to schedule recurring debts like allowances or shared rent. You can invite friends to share I.O.U.s with Facebook, Twitter or email, or you can choose to track debts privately.
As the designer and developer of this service, I'd love to get some feedback (and of course, early users). My company (Cerebris - I'm a co-founder) developed this application in order to try out a lot of tech, from Rails 3 to jQuery Mobile to Omniauth to hosting at Rackspace. Once we realized how useful it had become to us personally, we decided to add some extra polish and features.
We're keeping this service completely free, although we may add premium features in the future. We also plan to introduce native apps for a small fee (perhaps $1.99).
I really hope you find I.O.U. Mate fun and useful. Please let me know what you think - I'd appreciate feedback of any kind. I'm particularly interested to know how people like the new mobile interface.
Thanks in advance for checking it out!
http://IOUmate.com","title":"Ask HN: Please review my site IOUmate.com","updated_at":"2023-09-06T19:47:55Z","url":""},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jauco"},"title":{"matchLevel":"none","matchedWords":[],"value":"Orbit.js \u2013 A standalone library for data access and synchronization"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://github.com/cerebris/orbit.js"}},"_tags":["story","author_jauco","story_7086201"],"author":"jauco","children":[7086735,7086887,7087045,7087065],"created_at":"2014-01-19T21:59:18Z","created_at_i":1390168758,"num_comments":12,"objectID":"7086201","points":72,"story_id":7086201,"title":"Orbit.js \u2013 A standalone library for data access and synchronization","updated_at":"2023-09-06T21:57:32Z","url":"https://github.com/cerebris/orbit.js"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Beginning Ember.js on Rails: Part 2"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2012/01/26/beginning-ember-js-on-rails-part-2/"}},"_tags":["story","author_dgeb","story_3516061"],"author":"dgeb","children":[3516481],"created_at":"2012-01-26T21:22:33Z","created_at_i":1327612953,"num_comments":7,"objectID":"3516061","points":65,"story_id":3516061,"story_text":"","title":"Beginning Ember.js on Rails: Part 2","updated_at":"2023-09-06T21:02:31Z","url":"http://www.cerebris.com/blog/2012/01/26/beginning-ember-js-on-rails-part-2/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Beginning Ember.js on Rails: Part 3 (of 3)"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2012/01/31/beginning-ember-js-on-rails-part-3/"}},"_tags":["story","author_dgeb","story_3534894"],"author":"dgeb","children":[3534908,3534950],"created_at":"2012-01-31T19:56:00Z","created_at_i":1328039760,"num_comments":2,"objectID":"3534894","points":11,"story_id":3534894,"story_text":"","title":"Beginning Ember.js on Rails: Part 3 (of 3)","updated_at":"2023-09-06T21:03:55Z","url":"http://www.cerebris.com/blog/2012/01/31/beginning-ember-js-on-rails-part-3/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Beginning Ember.js on Rails: Part 1"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2012/01/24/beginning-ember-js-on-rails-part-1/"}},"_tags":["story","author_dgeb","story_3505488"],"author":"dgeb","created_at":"2012-01-24T16:46:42Z","created_at_i":1327423602,"num_comments":0,"objectID":"3505488","points":7,"story_id":3505488,"story_text":"","title":"Beginning Ember.js on Rails: Part 1","updated_at":"2024-09-19T18:12:50Z","url":"http://www.cerebris.com/blog/2012/01/24/beginning-ember-js-on-rails-part-1/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Understanding Ember.Object"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2012/03/06/understanding-ember-object/"}},"_tags":["story","author_dgeb","story_3671410"],"author":"dgeb","created_at":"2012-03-06T15:28:34Z","created_at_i":1331047714,"num_comments":0,"objectID":"3671410","points":5,"story_id":3671410,"story_text":"","title":"Understanding Ember.Object","updated_at":"2024-09-19T18:27:11Z","url":"http://www.cerebris.com/blog/2012/03/06/understanding-ember-object/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Warming up to Ember.js"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2012/01/20/warming-up-to-ember-js/"}},"_tags":["story","author_dgeb","story_3490855"],"author":"dgeb","created_at":"2012-01-20T19:35:20Z","created_at_i":1327088120,"num_comments":0,"objectID":"3490855","points":5,"story_id":3490855,"story_text":"","title":"Warming up to Ember.js","updated_at":"2024-09-19T18:18:24Z","url":"http://www.cerebris.com/blog/2012/01/20/warming-up-to-ember-js/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"daviducolo"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"JSON API 1.0"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2015/06/04/jsonapi-1-0/"}},"_tags":["story","author_daviducolo","story_9677850"],"author":"daviducolo","created_at":"2015-06-08T08:28:50Z","created_at_i":1433752130,"num_comments":0,"objectID":"9677850","points":3,"story_id":9677850,"story_text":"","title":"JSON API 1.0","updated_at":"2023-09-07T02:55:29Z","url":"http://www.cerebris.com/blog/2015/06/04/jsonapi-1-0/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Stop using the term \"Single Page Application\""},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2013/07/31/single-page-applications/"}},"_tags":["story","author_dgeb","story_6134693"],"author":"dgeb","created_at":"2013-07-31T16:01:15Z","created_at_i":1375286475,"num_comments":0,"objectID":"6134693","points":3,"story_id":6134693,"story_text":"","title":"Stop using the term \"Single Page Application\"","updated_at":"2024-09-19T19:48:46Z","url":"http://www.cerebris.com/blog/2013/07/31/single-page-applications/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Tame Rails Email Dragons with Mailhopper"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2011/09/07/tame-rails-email-dragons-with-mailhopper/"}},"_tags":["story","author_dgeb","story_2970757"],"author":"dgeb","created_at":"2011-09-07T18:29:44Z","created_at_i":1315420184,"num_comments":0,"objectID":"2970757","points":3,"story_id":2970757,"story_text":"","title":"Tame Rails Email Dragons with Mailhopper","updated_at":"2023-09-06T20:28:06Z","url":"http://www.cerebris.com/blog/2011/09/07/tame-rails-email-dragons-with-mailhopper/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"TodWhinch"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Provides a framework for developing a server that complies with JSON API"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://github.com/cerebris/jsonapi-resources"}},"_tags":["story","author_TodWhinch","story_9922860"],"author":"TodWhinch","created_at":"2015-07-21T15:24:52Z","created_at_i":1437492292,"num_comments":0,"objectID":"9922860","points":2,"story_id":9922860,"story_text":"","title":"Provides a framework for developing a server that complies with JSON API","updated_at":"2024-09-19T22:07:13Z","url":"https://github.com/cerebris/jsonapi-resources"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"JSON API reaches v1.0"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2015/06/04/jsonapi-1-0/"}},"_tags":["story","author_dgeb","story_9661034"],"author":"dgeb","created_at":"2015-06-04T18:16:19Z","created_at_i":1433441779,"num_comments":0,"objectID":"9661034","points":2,"story_id":9661034,"story_text":"","title":"JSON API reaches v1.0","updated_at":"2023-09-07T02:54:11Z","url":"http://www.cerebris.com/blog/2015/06/04/jsonapi-1-0/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Jsonapi-resources \u2013 Easily Support JSON API in Rails"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2014/08/22/introducing-jsonapi-resources/"}},"_tags":["story","author_dgeb","story_8213722"],"author":"dgeb","created_at":"2014-08-22T20:11:31Z","created_at_i":1408738291,"num_comments":0,"objectID":"8213722","points":2,"story_id":8213722,"story_text":"","title":"Jsonapi-resources \u2013 Easily Support JSON API in Rails","updated_at":"2024-09-19T21:03:30Z","url":"http://www.cerebris.com/blog/2014/08/22/introducing-jsonapi-resources/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"The \"Development Drawbacks\" of JavaScript Web Applications"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2013/08/08/the-development-drawbacks-of-javascript-web-applications/"}},"_tags":["story","author_dgeb","story_6183547"],"author":"dgeb","created_at":"2013-08-09T03:18:04Z","created_at_i":1376018284,"num_comments":0,"objectID":"6183547","points":2,"story_id":6183547,"story_text":"","title":"The \"Development Drawbacks\" of JavaScript Web Applications","updated_at":"2023-09-06T21:55:54Z","url":"http://www.cerebris.com/blog/2013/08/08/the-development-drawbacks-of-javascript-web-applications/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dgeb"},"story_text":{"matchLevel":"none","matchedWords":[],"value":""},"title":{"matchLevel":"none","matchedWords":[],"value":"Thought experiment for deciding a web app's architecture"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"http://www.cerebris.com/blog/2014/01/14/of-satellites-and-rovers/"}},"_tags":["story","author_dgeb","story_7061314"],"author":"dgeb","created_at":"2014-01-15T02:34:22Z","created_at_i":1389753262,"num_comments":0,"objectID":"7061314","points":1,"story_id":7061314,"story_text":"","title":"Thought experiment for deciding a web app's architecture","updated_at":"2023-09-06T21:55:23Z","url":"http://www.cerebris.com/blog/2014/01/14/of-satellites-and-rovers/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"asb"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras-GPT: A Family of Open, Compute-Efficient, Large Language Models"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.net/blog/cerebras-gpt-a-family-of-open-compute-efficient-large-language-models/"}},"_tags":["story","author_asb","story_35343763"],"author":"asb","children":[35344068,35344112,35344166,35344272,35344428,35344498,35344536,35344569,35344626,35344706,35344737,35344760,35344932,35345247,35345296,35345361,35345415,35345731,35345781,35345836,35345980,35346328,35346357,35346685,35348409,35348609,35348782,35349244,35350379,35372506],"created_at":"2023-03-28T16:34:15Z","created_at_i":1680021255,"num_comments":232,"objectID":"35343763","points":567,"story_id":35343763,"title":"Cerebras-GPT: A Family of Open, Compute-Efficient, Large Language Models","updated_at":"2025-10-06T20:43:11Z","url":"https://www.cerebras.net/blog/cerebras-gpt-a-family-of-open-compute-efficient-large-language-models/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"d3vr"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Code"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/blog/introducing-cerebras-code"}},"_tags":["story","author_d3vr","story_44762959"],"author":"d3vr","children":[44763172,44763219,44763270,44763281,44763317,44763339,44763364,44763369,44763376,44763392,44763436,44763437,44763508,44763628,44763634,44763722,44763760,44763808,44764007,44764209,44764365,44764378,44764497,44764610,44764671,44764759,44764832,44765201,44765331,44765725,44766555,44766650,44766706,44767096,44768089,44768249,44768368],"created_at":"2025-08-01T22:04:38Z","created_at_i":1754085878,"num_comments":172,"objectID":"44762959","points":449,"story_id":44762959,"title":"Cerebras Code","updated_at":"2026-07-02T00:30:33Z","url":"https://www.cerebras.ai/blog/introducing-cerebras-code"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"benchmarkist"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Llama 3.1 405B now runs at 969 tokens/s on Cerebras Inference"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://cerebras.ai/blog/llama-405b-inference"}},"_tags":["story","author_benchmarkist","story_42178761"],"author":"benchmarkist","children":[42179444,42179476,42179482,42179527,42179539,42179727,42179744,42179843,42179871,42179943,42179970,42180121,42180200,42180257,42180484,42180550,42180575,42180666,42180721,42181071,42181139,42181280,42181295,42181543,42181951,42182024,42182449,42183087,42232579],"created_at":"2024-11-19T00:15:04Z","created_at_i":1731975304,"num_comments":156,"objectID":"42178761","points":427,"story_id":42178761,"title":"Llama 3.1 405B now runs at 969 tokens/s on Cerebras Inference","updated_at":"2026-04-24T07:48:29Z","url":"https://cerebras.ai/blog/llama-405b-inference"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"mihau"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras launches Qwen3-235B, achieving 1.5k tokens per second"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/press-release/cerebras-launches-qwen3-235b-world-s-fastest-frontier-ai-model-with-full-131k-context-support"}},"_tags":["story","author_mihau","story_44657727"],"author":"mihau","children":[44657982,44657987,44658006,44658040,44658077,44658114,44658140,44658173,44658198,44658288,44658337,44658483,44658571,44658610,44658775,44658822,44658862,44659318,44659423,44659470,44659526,44659530,44659556,44660084,44660127,44660605,44660704,44660709,44660710,44662403,44662962,44663272,44663398,44664749],"created_at":"2025-07-23T10:51:51Z","created_at_i":1753267911,"num_comments":155,"objectID":"44657727","points":364,"story_id":44657727,"title":"Cerebras launches Qwen3-235B, achieving 1.5k tokens per second","updated_at":"2026-01-05T20:26:54Z","url":"https://www.cerebras.ai/press-release/cerebras-launches-qwen3-235b-world-s-fastest-frontier-ai-model-with-full-131k-context-support"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"robg"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"First images of cerebrospinal fluid in the brain during sleep (2019)"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.bu.edu/articles/2019/cerebrospinal-fluid-washing-in-brain-during-sleep/"}},"_tags":["story","author_robg","story_30222210"],"author":"robg","children":[30222766,30222813,30222975,30223104,30223152,30223171,30223397,30223529,30223551,30223624,30224191,30224665,30225235,30225471,30225557,30225811,30226529,30228319,30228610,30236362],"created_at":"2022-02-05T15:49:25Z","created_at_i":1644076165,"num_comments":111,"objectID":"30222210","points":347,"story_id":30222210,"title":"First images of cerebrospinal fluid in the brain during sleep (2019)","updated_at":"2025-03-30T06:51:08Z","url":"https://www.bu.edu/articles/2019/cerebrospinal-fluid-washing-in-brain-during-sleep/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"Anon84"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras\u2019 new monster AI chip adds 1.4T transistors"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://spectrum.ieee.org/tech-talk/semiconductors/processors/cerebras-giant-ai-chip-now-has-a-trillions-more-transistors"}},"_tags":["story","author_Anon84","story_26907176"],"author":"Anon84","children":[26907709,26907743,26907848,26907894,26907956,26908087,26908174,26908207,26908221,26908335,26908407,26908455,26908458,26908530,26908679,26908741,26908867,26908990,26909055,26909113,26909206,26909575,26909620,26909745,26909778,26909938,26911224,26913498],"created_at":"2021-04-22T19:37:25Z","created_at_i":1619120245,"num_comments":165,"objectID":"26907176","points":273,"story_id":26907176,"title":"Cerebras\u2019 new monster AI chip adds 1.4T transistors","updated_at":"2024-09-20T08:22:15Z","url":"https://spectrum.ieee.org/tech-talk/semiconductors/processors/cerebras-giant-ai-chip-now-has-a-trillions-more-transistors"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"freeqaz"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras-GPT vs. LLaMA AI Model Performance Comparison"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.lunasec.io/docs/blog/cerebras-gpt-vs-llama-ai-model-comparison/"}},"_tags":["story","author_freeqaz","story_35362048"],"author":"freeqaz","children":[35363629,35363651,35363769,35364446,35365548,35366314,35366739],"created_at":"2023-03-29T19:26:17Z","created_at_i":1680117977,"num_comments":70,"objectID":"35362048","points":263,"story_id":35362048,"title":"Cerebras-GPT vs. LLaMA AI Model Performance Comparison","updated_at":"2025-10-06T20:43:11Z","url":"https://www.lunasec.io/docs/blog/cerebras-gpt-vs-llama-ai-model-comparison/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"modeless"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Systems unveils a 1.2T transistor chip for AI"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://venturebeat.com/2019/08/19/cerebras-systems-unveils-a-record-1-2-trillion-transistor-chip-for-ai/"}},"_tags":["story","author_modeless","story_20739408"],"author":"modeless","children":[20739579,20739639,20739650,20739666,20739667,20739677,20739717,20739733,20739736,20739751,20739770,20739771,20739825,20739827,20739830,20739878,20739905,20739951,20740046,20740272,20740329,20740653,20740813,20740833,20741278,20741521,20741579,20741753,20741775,20742918],"created_at":"2019-08-19T16:52:44Z","created_at_i":1566233564,"num_comments":263,"objectID":"20739408","points":227,"story_id":20739408,"title":"Cerebras Systems unveils a 1.2T transistor chip for AI","updated_at":"2025-10-13T10:05:58Z","url":"https://venturebeat.com/2019/08/19/cerebras-systems-unveils-a-record-1-2-trillion-transistor-chip-for-ai/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"TradingPlaces"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"AI chipmaker Cerebras files for IPO"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cnbc.com/2024/09/30/cerebras-files-for-ipo.html"}},"_tags":["story","author_TradingPlaces","story_41702789"],"author":"TradingPlaces","children":[41703113,41703152,41703153,41703214,41703346,41703363,41703716,41703867,41703905,41704010,41704031,41704240,41704391,41704617,41704726,41704769,41705249,41705745,41706254,41706409,41707845,41711448,41718662],"created_at":"2024-09-30T22:26:49Z","created_at_i":1727735209,"num_comments":135,"objectID":"41702789","points":219,"story_id":41702789,"title":"AI chipmaker Cerebras files for IPO","updated_at":"2026-04-19T13:27:11Z","url":"https://www.cnbc.com/2024/09/30/cerebras-files-for-ipo.html"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"nathabonfim59"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Code now supports GLM 4.6 at 1000 tokens/sec"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/code"}},"_tags":["story","author_nathabonfim59","story_45852751"],"author":"nathabonfim59","children":[45853884,45853933,45853998,45854017,45854033,45854075,45854134,45854166,45854205,45854280,45854543,45854577,45854770,45854885,45855742,45855844,45856628,45856663,45856787,45856984,45857130,45857403,45858487,45858718,45859014,45863100,45865180,45867635,46001451],"created_at":"2025-11-08T00:00:27Z","created_at_i":1762560027,"num_comments":129,"objectID":"45852751","points":194,"story_id":45852751,"title":"Cerebras Code now supports GLM 4.6 at 1000 tokens/sec","updated_at":"2026-03-05T22:59:26Z","url":"https://www.cerebras.ai/code"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"meetpateltech"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Inference: AI at Instant Speed"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://cerebras.ai/blog/introducing-cerebras-inference-ai-at-instant-speed"}},"_tags":["story","author_meetpateltech","story_41369705"],"author":"meetpateltech","children":[41370166,41370207,41370253,41370265,41370306,41370308,41370339,41370374,41370637,41370697,41371041,41371389,41372903,41377042,41377063],"created_at":"2024-08-27T16:49:46Z","created_at_i":1724777386,"num_comments":72,"objectID":"41369705","points":174,"story_id":41369705,"title":"Cerebras Inference: AI at Instant Speed","updated_at":"2025-10-15T20:35:20Z","url":"https://cerebras.ai/blog/introducing-cerebras-inference-ai-at-instant-speed"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"pheme1"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras\u2019s giant chip will smash deep learning\u2019s speed barrier"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://spectrum.ieee.org/semiconductors/processors/cerebrass-giant-chip-will-smash-deep-learnings-speed-barrier"}},"_tags":["story","author_pheme1","story_21937429"],"author":"pheme1","children":[21937688,21938017,21938163,21938180,21938620,21938689,21938951,21939016,21939491,21940023,21940126,21940783,21940929,21943219],"created_at":"2020-01-02T17:26:13Z","created_at_i":1577985973,"num_comments":102,"objectID":"21937429","points":159,"story_id":21937429,"title":"Cerebras\u2019s giant chip will smash deep learning\u2019s speed barrier","updated_at":"2024-09-20T05:30:08Z","url":"https://spectrum.ieee.org/semiconductors/processors/cerebrass-giant-chip-will-smash-deep-learnings-speed-barrier"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"campers"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Inference now 3x faster: Llama3.1-70B breaks 2,100 tokens/s"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://cerebras.ai/blog/cerebras-inference-3x-faster"}},"_tags":["story","author_campers","story_41941883"],"author":"campers","children":[41942463,41942469,41942477,41942508,41942582,41942623,41942657,41943037,41943160,41943256,41943277,41943395,41943442,41944087,41945906],"created_at":"2024-10-25T03:04:52Z","created_at_i":1729825492,"num_comments":84,"objectID":"41941883","points":147,"story_id":41941883,"title":"Cerebras Inference now 3x faster: Llama3.1-70B breaks 2,100 tokens/s","updated_at":"2024-12-22T04:06:26Z","url":"https://cerebras.ai/blog/cerebras-inference-3x-faster"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"unwind"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"The Cerebras CS-2 wafer-scale engine (850k cores, 40 GB SRAM)"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://cerebras.net/product/"}},"_tags":["story","author_unwind","story_27459466"],"author":"unwind","children":[27460283,27460290,27460291,27460379,27460667,27460825,27460875,27460916,27461596,27461703,27461810,27462099,27462853],"created_at":"2021-06-10T11:34:03Z","created_at_i":1623324843,"num_comments":52,"objectID":"27459466","points":127,"story_id":27459466,"title":"The Cerebras CS-2 wafer-scale engine (850k cores, 40 GB SRAM)","updated_at":"2024-09-20T08:45:35Z","url":"https://cerebras.net/product/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"fcpguru"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras systems raises $1.1B Series G"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/press-release/series-g"}},"_tags":["story","author_fcpguru","story_45427111"],"author":"fcpguru","children":[45427129,45428878,45428894,45429098,45429179,45429212,45429248,45429316,45429434,45429514,45429569,45429924,45430226,45430662,45430697,45433192,45442636],"created_at":"2025-09-30T15:54:14Z","created_at_i":1759247654,"num_comments":75,"objectID":"45427111","points":126,"story_id":45427111,"title":"Cerebras systems raises $1.1B Series G","updated_at":"2026-06-24T19:18:05Z","url":"https://www.cerebras.ai/press-release/series-g"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"asdfasdf1"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"4T transistors, one giant chip (Cerebras WSE-3) [video]"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.youtube.com/watch?v=f4Dly8I8lMY"}},"_tags":["story","author_asdfasdf1","story_39693356"],"author":"asdfasdf1","children":[39693357,39693748,39693769,39693930,39693948,39694060,39694106,39694114,39694123,39694467,39694481,39694924,39695134,39695136,39695446,39695538,39697133,39697764,39697778,39697788,39697842,39697976,39697984,39698068,39698147,39698157,39698159],"created_at":"2024-03-13T16:13:36Z","created_at_i":1710346416,"num_comments":86,"objectID":"39693356","points":118,"story_id":39693356,"title":"4T transistors, one giant chip (Cerebras WSE-3) [video]","updated_at":"2024-09-20T16:42:05Z","url":"https://www.youtube.com/watch?v=f4Dly8I8lMY"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"cs-fan-101"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras-GPT: Open Compute-Optimal Language Models Trained on Cerebras Cluster"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://arxiv.org/abs/2304.03208"}},"_tags":["story","author_cs-fan-101","story_35487846"],"author":"cs-fan-101","children":[35487847,35489765,35490837,35491614,35492003],"created_at":"2023-04-07T22:26:38Z","created_at_i":1680906398,"num_comments":12,"objectID":"35487846","points":97,"story_id":35487846,"title":"Cerebras-GPT: Open Compute-Optimal Language Models Trained on Cerebras Cluster","updated_at":"2024-09-20T13:52:49Z","url":"https://arxiv.org/abs/2304.03208"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"_micah_h"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras launches inference for Llama 3.1; benchmarked at 1846 tokens/s on 8B"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://twitter.com/ArtificialAnlys/status/1828463912389402896"}},"_tags":["story","author__micah_h","story_41369586"],"author":"_micah_h","children":[41370183,41370590,41370906,41371216,41371365,41372221,41372519,41373405,41374188],"created_at":"2024-08-27T16:42:01Z","created_at_i":1724776921,"num_comments":42,"objectID":"41369586","points":95,"story_id":41369586,"title":"Cerebras launches inference for Llama 3.1; benchmarked at 1846 tokens/s on 8B","updated_at":"2024-09-20T17:41:53Z","url":"https://twitter.com/ArtificialAnlys/status/1828463912389402896"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ByteAtATime"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras achieves 2,500T/s on Llama 4 Maverick (400B)"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/press-release/maverick"}},"_tags":["story","author_ByteAtATime","story_44141728"],"author":"ByteAtATime","children":[44142361,44142442,44142530,44142571,44142679,44142892,44142921,44143256,44143292,44143630,44143830,44148021],"created_at":"2025-05-31T03:49:22Z","created_at_i":1748663362,"num_comments":93,"objectID":"44141728","points":93,"story_id":44141728,"title":"Cerebras achieves 2,500T/s on Llama 4 Maverick (400B)","updated_at":"2026-06-16T22:49:51Z","url":"https://www.cerebras.ai/press-release/maverick"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"gmays"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Nasopharyngeal Lymphatics Found to Be Crucial for Cerebrospinal Fluid Outflow"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.ibs.re.kr/cop/bbs/BBSMSTR_000000000738/selectBoardArticle.do?nttId=24483&pageIndex=1&searchCnd=&searchWrd="}},"_tags":["story","author_gmays","story_39219953"],"author":"gmays","children":[39221183,39221188,39221199,39221567,39221937,39222228,39222506,39222617,39224018,39224622,39225344,39225596],"created_at":"2024-02-01T19:17:16Z","created_at_i":1706815036,"num_comments":38,"objectID":"39219953","points":93,"story_id":39219953,"title":"Nasopharyngeal Lymphatics Found to Be Crucial for Cerebrospinal Fluid Outflow","updated_at":"2024-09-20T16:14:02Z","url":"https://www.ibs.re.kr/cop/bbs/BBSMSTR_000000000738/selectBoardArticle.do?nttId=24483&pageIndex=1&searchCnd=&searchWrd="},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"rbanffy"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"A Look at Cerebras Wafer-Scale Engine: Half Square Foot Silicon Chip"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://fuse.wikichip.org/news/3010/a-look-at-cerebras-wafer-scale-engine-half-square-foot-silicon-chip/"}},"_tags":["story","author_rbanffy","story_21556472"],"author":"rbanffy","children":[21556675,21556734,21556738,21556820,21556888,21556905,21557241,21557511,21557653,21558322,21558453,21558794],"created_at":"2019-11-17T03:55:22Z","created_at_i":1573962922,"num_comments":59,"objectID":"21556472","points":88,"story_id":21556472,"title":"A Look at Cerebras Wafer-Scale Engine: Half Square Foot Silicon Chip","updated_at":"2024-09-20T05:10:41Z","url":"https://fuse.wikichip.org/news/3010/a-look-at-cerebras-wafer-scale-engine-half-square-foot-silicon-chip/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ramshanker"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Wafer Scale Engine Gen2 7nm 2.6T Transistors"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.servethehome.com/cerebras-wafer-scale-engine-gen2-7nm-2-6t-transistors/"}},"_tags":["story","author_ramshanker","story_24206308"],"author":"ramshanker","children":[24206331,24206524,24206549,24206767,24206786,24206834,24206991,24207035,24207037,24207130,24207581,24207892,24208132,24208239,24208374],"created_at":"2020-08-19T01:35:12Z","created_at_i":1597800912,"num_comments":54,"objectID":"24206308","points":66,"story_id":24206308,"title":"Cerebras Wafer Scale Engine Gen2 7nm 2.6T Transistors","updated_at":"2024-09-20T06:43:11Z","url":"https://www.servethehome.com/cerebras-wafer-scale-engine-gen2-7nm-2-6t-transistors/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"rbanffy"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Trains Llama Models to Leap over GPUs"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.nextplatform.com/2024/10/25/cerebras-trains-llama-models-to-leap-over-gpus/"}},"_tags":["story","author_rbanffy","story_41956213"],"author":"rbanffy","children":[42002870,42002904,42002921,42003170,42003352,42003656,42003688,42005262,42005913],"created_at":"2024-10-26T17:26:02Z","created_at_i":1729963562,"num_comments":33,"objectID":"41956213","points":64,"story_id":41956213,"title":"Cerebras Trains Llama Models to Leap over GPUs","updated_at":"2025-08-25T12:21:47Z","url":"https://www.nextplatform.com/2024/10/25/cerebras-trains-llama-models-to-leap-over-gpus/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"za_mike157"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Hi HN,
We are Michael & Jono, and we are building Cerebrium (https://www.cerebrium.ai), a serverless infrastructure platform for ML/AI applications - we make it easy for engineers to build, deploy and scale AI applications.
Initially, we\u2019ve been hyper-focused on the inference side of applications, but we\u2019re working on expanding our functionality to support training and data processing use cases\u2014eventually covering the full AI development lifecycle.
You can watch a quick loom video of us deploying: https://www.loom.com/share/06947794b3bf4bb1bb21c87066dfcc66?...
How we got here:
Jono and I led the technical team at our previous e-commerce startup, which grew rapidly over a few years. As we scaled, we were tasked with building out ML applications to make the business more efficient. It was tough\u2014every day felt like a defeat. We found ourselves stitching together AWS Lambda, Sagemaker, and Prefect jobs (this stack alone was enough to make me want to give up). By the time we reached production, the costs were too high to maintain. Getting these applications live required a significant upfront investment of both time and money, making it inaccessible for most startups and scale-ups to attempt. We wanted to create something that would help us (and others like us) implement ML/AI applications easily and cost-effectively.
The problem:
There are a ton of challenges to tackle to realize our vision, but we\u2019ve initially focused on a few key ones:
1. GPUs are expensive \u2013 An A100 is 326 times the cost of a CPU, and companies are using LLMs like they\u2019re simple APIs. Serverless instances solve this to an extent, but minimizing cold starts is difficult.
2. Local development \u2013 Engineers need local development environments to iterate quickly, but production-grade GPUs aren\u2019t available on consumer hardware. How can we make cloud deployments feel as fast as just saving a file locally and retrying?
3. Cost to experiment - To run experiments we had to spin up EC2 instances each day, recreate our environment and run scripts. It was difficult to monitor logs, instance usage metrics as well as run large processing jobs or scale endpoints without a significant infrastructure investment. Additionally, we often forget to switch off instances which cost us money!
Our Approach
We have three core areas that we are focused on which we believe are the most important for any infrastructure platform:
1. Performance:
We have worked hard to get our added network latency <50ms and the cold start of our average workloads to 2-4 seconds. Here are a few things we did to get our cold starts so low:
- Container Runtime: We built our own container runtime that splits container images into two parts\u2014metadata and data blobs. Metadata provides the file structure, while the actual data blobs are fetched on-demand. This allows containers to start before the full image is downloaded. In the background, we prefetch the remaining blobs.
- Caching: Once an image is on a machine, it\u2019s cached for future use. This makes subsequent container startups much faster. We also intelligently route requests to machines where the image is already cached.
- Efficient Inference: We route requests to the optimal machines, prioritizing low-latency and high-throughput performance. If no containers are immediately available, we efficiently queue the requests through our task scheduling system.
- Distributed Storage Cache: One of the most resource-intensive parts of AI workloads is loading models into VRAM. We use NVME drives (which are much faster than network volumes), as close as possible to the machines and we orchestrate workloads to nodes that already contain the necessary model weights where possible.
2. Developer Experience
We built Cerebrium to help developers iterate as quickly as possible by streamlining the entire build and deployment process.
To get build times as low as possible, we use high-performance machines and cache layers where possible. We've reduced first-time build times to an average of 2 minutes and 24 seconds, with subsequent builds completing in just 19 seconds.
We also offer a wide range of GPU types\u2014over 8 different options\u2014so you can easily test performance and cost efficiency by adjusting a single line in your configuration file.
To reduce friction, we\u2019ve kept things simple. There are no custom Python decorators, no Cerebrium specific syntax to learn. You just add a .toml file to define your hardware requirements and environment settings. This makes migrating onto or off our platform just as easy as migrating off. We aim to impress you enough that you will want to stay.
3. Stability
This is arguably more important than the first two areas - no one wants to get an email at 11pm at night or on a Saturday that their application is down or degraded. Since April, we\u2019ve maintained 99.999% uptime. We have redundancies in place, monitoring, alerts, and a team that covers all time zones to resolve any issues quickly.
Why Is This Hard?
Building Cerebrium has been challenging because it involves solving multiple interconnected problems. It requires optimization at every step\u2014from efficiently splitting images to fetching data on-demand without introducing latency, handling distributed caching, optimizing our network stack, and ensuring redundancies, all while holding true to the three areas mentioned above.
Pricing:
We charge you exactly for the resources you need and only charge you when your code is running ie: usage-based. For example, if you specify you need 1 A100 GPU, with 2 CPUs and 12 GB of RAM we charge you exactly for that and not a full A100 (12 CPU\u2019s and 148GB of memory)
You can see more about our pricing here: http://www.cerebrium.ai/pricing
What\u2019s Next?
We're builders too, and we know how crucial support can be when you're working on something new. Here's what we've put together to support teams like yours:
- $30 in free credit to start exploring. If you're onto something interesting but need more runway, just give us a shout - we\u2019d be happy to extend that for compelling use-cases.
- We have worked hard on our docs to make onboarding easy as well as have a very elaborate Github repo covering AI voice agents, LLM optimizations and much more.
Docs: https://docs.cerebrium.ai/cerebrium/getting-started/introduc...
Github Examples: https://github.com/CerebriumAI/examples/tree/master
If you have a question or hit a snag, you can directly reach out to the engineers who built the platform\u2014we\u2019re here to help! We\u2019ve also set up Slack and Discord communities where you can connect with other creators, share experiences, ask for advice, or just chat with folks building cool things.
We're looking forward to seeing what you all build and please give us feedback on what you would like us to improve/add"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Launch HN: Cerebrium (YC W22) \u2013 Serverless Infrastructure Platform for ML/AI"}},"_tags":["story","author_za_mike157","story_41579777","launch_hn"],"author":"za_mike157","children":[41581063,41581606,41581641,41581978,41582412,41583034,41583567,41585296,41585638,41586731,41588056,41588178,41589141],"created_at":"2024-09-18T13:54:54Z","created_at_i":1726667694,"num_comments":32,"objectID":"41579777","points":48,"story_id":41579777,"story_text":"Hi HN,
We are Michael & Jono, and we are building Cerebrium (https://www.cerebrium.ai), a serverless infrastructure platform for ML/AI applications - we make it easy for engineers to build, deploy and scale AI applications.
Initially, we\u2019ve been hyper-focused on the inference side of applications, but we\u2019re working on expanding our functionality to support training and data processing use cases\u2014eventually covering the full AI development lifecycle.
You can watch a quick loom video of us deploying: https://www.loom.com/share/06947794b3bf4bb1bb21c87066dfcc66?...
How we got here:
Jono and I led the technical team at our previous e-commerce startup, which grew rapidly over a few years. As we scaled, we were tasked with building out ML applications to make the business more efficient. It was tough\u2014every day felt like a defeat. We found ourselves stitching together AWS Lambda, Sagemaker, and Prefect jobs (this stack alone was enough to make me want to give up). By the time we reached production, the costs were too high to maintain. Getting these applications live required a significant upfront investment of both time and money, making it inaccessible for most startups and scale-ups to attempt. We wanted to create something that would help us (and others like us) implement ML/AI applications easily and cost-effectively.
The problem:
There are a ton of challenges to tackle to realize our vision, but we\u2019ve initially focused on a few key ones:
1. GPUs are expensive \u2013 An A100 is 326 times the cost of a CPU, and companies are using LLMs like they\u2019re simple APIs. Serverless instances solve this to an extent, but minimizing cold starts is difficult.
2. Local development \u2013 Engineers need local development environments to iterate quickly, but production-grade GPUs aren\u2019t available on consumer hardware. How can we make cloud deployments feel as fast as just saving a file locally and retrying?
3. Cost to experiment - To run experiments we had to spin up EC2 instances each day, recreate our environment and run scripts. It was difficult to monitor logs, instance usage metrics as well as run large processing jobs or scale endpoints without a significant infrastructure investment. Additionally, we often forget to switch off instances which cost us money!
Our Approach
We have three core areas that we are focused on which we believe are the most important for any infrastructure platform:
1. Performance:
We have worked hard to get our added network latency <50ms and the cold start of our average workloads to 2-4 seconds. Here are a few things we did to get our cold starts so low:
- Container Runtime: We built our own container runtime that splits container images into two parts\u2014metadata and data blobs. Metadata provides the file structure, while the actual data blobs are fetched on-demand. This allows containers to start before the full image is downloaded. In the background, we prefetch the remaining blobs.
- Caching: Once an image is on a machine, it\u2019s cached for future use. This makes subsequent container startups much faster. We also intelligently route requests to machines where the image is already cached.
- Efficient Inference: We route requests to the optimal machines, prioritizing low-latency and high-throughput performance. If no containers are immediately available, we efficiently queue the requests through our task scheduling system.
- Distributed Storage Cache: One of the most resource-intensive parts of AI workloads is loading models into VRAM. We use NVME drives (which are much faster than network volumes), as close as possible to the machines and we orchestrate workloads to nodes that already contain the necessary model weights where possible.
2. Developer Experience
We built Cerebrium to help developers iterate as quickly as possible by streamlining the entire build and deployment process.
To get build times as low as possible, we use high-performance machines and cache layers where possible. We've reduced first-time build times to an average of 2 minutes and 24 seconds, with subsequent builds completing in just 19 seconds.
We also offer a wide range of GPU types\u2014over 8 different options\u2014so you can easily test performance and cost efficiency by adjusting a single line in your configuration file.
To reduce friction, we\u2019ve kept things simple. There are no custom Python decorators, no Cerebrium specific syntax to learn. You just add a .toml file to define your hardware requirements and environment settings. This makes migrating onto or off our platform just as easy as migrating off. We aim to impress you enough that you will want to stay.
3. Stability
This is arguably more important than the first two areas - no one wants to get an email at 11pm at night or on a Saturday that their application is down or degraded. Since April, we\u2019ve maintained 99.999% uptime. We have redundancies in place, monitoring, alerts, and a team that covers all time zones to resolve any issues quickly.
Why Is This Hard?
Building Cerebrium has been challenging because it involves solving multiple interconnected problems. It requires optimization at every step\u2014from efficiently splitting images to fetching data on-demand without introducing latency, handling distributed caching, optimizing our network stack, and ensuring redundancies, all while holding true to the three areas mentioned above.
Pricing:
We charge you exactly for the resources you need and only charge you when your code is running ie: usage-based. For example, if you specify you need 1 A100 GPU, with 2 CPUs and 12 GB of RAM we charge you exactly for that and not a full A100 (12 CPU\u2019s and 148GB of memory)
You can see more about our pricing here: http://www.cerebrium.ai/pricing
What\u2019s Next?
We're builders too, and we know how crucial support can be when you're working on something new. Here's what we've put together to support teams like yours:
- $30 in free credit to start exploring. If you're onto something interesting but need more runway, just give us a shout - we\u2019d be happy to extend that for compelling use-cases.
- We have worked hard on our docs to make onboarding easy as well as have a very elaborate Github repo covering AI voice agents, LLM optimizations and much more.
Docs: https://docs.cerebrium.ai/cerebrium/getting-started/introduc...
Github Examples: https://github.com/CerebriumAI/examples/tree/master
If you have a question or hit a snag, you can directly reach out to the engineers who built the platform\u2014we\u2019re here to help! We\u2019ve also set up Slack and Discord communities where you can connect with other creators, share experiences, ask for advice, or just chat with folks building cool things.
We're looking forward to seeing what you all build and please give us feedback on what you would like us to improve/add","title":"Launch HN: Cerebrium (YC W22) \u2013 Serverless Infrastructure Platform for ML/AI","updated_at":"2024-10-05T21:16:13Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"samspenc"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"GPT-OSS 120B Runs at 3000 tokens/sec on Cerebras"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/blog/openai-gpt-oss-120b-runs-fastest-on-cerebras"}},"_tags":["story","author_samspenc","story_45853849"],"author":"samspenc","children":[45855692,45855949,45855962,45856279,45859802,45863329],"created_at":"2025-11-08T03:24:25Z","created_at_i":1762572265,"num_comments":29,"objectID":"45853849","points":48,"story_id":45853849,"title":"GPT-OSS 120B Runs at 3000 tokens/sec on Cerebras","updated_at":"2026-03-05T22:59:30Z","url":"https://www.cerebras.ai/blog/openai-gpt-oss-120b-runs-fastest-on-cerebras"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"sixhobbits"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"How to Migrate from OpenAI to Cerebrium for Cost-Predictable AI Inference"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://ritza.co/articles/migrate-from-openai-to-cerebrium-with-vllm-for-predictable-inference/"}},"_tags":["story","author_sixhobbits","story_44644404"],"author":"sixhobbits","children":[44644469,44644516,44644792,44644817,44644864,44644868,44645003,44645235,44645260,44650366],"created_at":"2025-07-22T08:08:37Z","created_at_i":1753171717,"num_comments":29,"objectID":"44644404","points":48,"story_id":44644404,"title":"How to Migrate from OpenAI to Cerebrium for Cost-Predictable AI Inference","updated_at":"2025-10-20T20:35:36Z","url":"https://ritza.co/articles/migrate-from-openai-to-cerebrium-with-vllm-for-predictable-inference/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"retreatguru"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Qwen3 Coder 480B is Live on Cerebras"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/blog/qwen3-coder-480b-is-live-on-cerebras"}},"_tags":["story","author_retreatguru","story_44760023"],"author":"retreatguru","children":[44760066,44761162,44762862,44763051,44763221,44769820],"created_at":"2025-08-01T17:50:19Z","created_at_i":1754070619,"num_comments":10,"objectID":"44760023","points":47,"story_id":44760023,"title":"Qwen3 Coder 480B is Live on Cerebras","updated_at":"2025-08-11T20:54:11Z","url":"https://www.cerebras.ai/blog/qwen3-coder-480b-is-live-on-cerebras"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"herpderperator"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras S-1"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.sec.gov/Archives/edgar/data/2021728/000162828026025762/cerebras-sx1april2026.htm"}},"_tags":["story","author_herpderperator","story_47810357"],"author":"herpderperator","children":[47811121,47811953,47813948],"created_at":"2026-04-17T20:42:19Z","created_at_i":1776458539,"num_comments":9,"objectID":"47810357","points":40,"story_id":47810357,"title":"Cerebras S-1","updated_at":"2026-05-16T07:48:01Z","url":"https://www.sec.gov/Archives/edgar/data/2021728/000162828026025762/cerebras-sx1april2026.htm"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"rbanffy"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras Unveils Wafer Scale Engine Two (WSE2): 2.6T Transistors, 100% Yield"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.anandtech.com/show/16626/cerebras-unveils-wafer-scale-engine-two-wse2-26-trillion-transistors-100-yield"}},"_tags":["story","author_rbanffy","story_26881514"],"author":"rbanffy","children":[26883697,26884449,26885101,26887393],"created_at":"2021-04-20T20:53:57Z","created_at_i":1618952037,"num_comments":15,"objectID":"26881514","points":38,"story_id":26881514,"title":"Cerebras Unveils Wafer Scale Engine Two (WSE2): 2.6T Transistors, 100% Yield","updated_at":"2024-09-20T08:26:35Z","url":"https://www.anandtech.com/show/16626/cerebras-unveils-wafer-scale-engine-two-wse2-26-trillion-transistors-100-yield"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"remusomega"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"I was an Enterprise customer on their platform. Cerebras began terminating production models and replacing them every few months. This time they notified us they are terminating Llama 3.3 70B, which is the model my plan was subscribed to.
Instead of offering us an alternative plan they told us all others are sold out and have kicked us off their platform.
Their Discord support group has multiple Enterprise customers all getting the same treatment. Their own support staff is telling them to migrate to Groq.
You can\u2019t build a stable business with a company that randomly terminates models this frequently. Who does not respect their customers (especially Enterprise customers, not even small utilizers), offering zero contingency plans if they decide to axe your model.
Because of their architecture, they can only host a finite number of models. Turnover is fast, so if your model gets marked for deprecation, you will get forced off the platform.
Cerebras is cool for personal Hobby projects, but is in absolutely no position to be selling "Enterprise" accounts to businesses."},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Tell HN: Avoid Cerebras if you are a founder"}},"_tags":["story","author_remusomega","story_46707904","ask_hn"],"author":"remusomega","children":[46707992,46709069,46710129,46710880,46716090],"created_at":"2026-01-21T16:29:09Z","created_at_i":1769012949,"num_comments":14,"objectID":"46707904","points":35,"story_id":46707904,"story_text":"I was an Enterprise customer on their platform. Cerebras began terminating production models and replacing them every few months. This time they notified us they are terminating Llama 3.3 70B, which is the model my plan was subscribed to.
Instead of offering us an alternative plan they told us all others are sold out and have kicked us off their platform.
Their Discord support group has multiple Enterprise customers all getting the same treatment. Their own support staff is telling them to migrate to Groq.
You can\u2019t build a stable business with a company that randomly terminates models this frequently. Who does not respect their customers (especially Enterprise customers, not even small utilizers), offering zero contingency plans if they decide to axe your model.
Because of their architecture, they can only host a finite number of models. Turnover is fast, so if your model gets marked for deprecation, you will get forced off the platform.
Cerebras is cool for personal Hobby projects, but is in absolutely no position to be selling "Enterprise" accounts to businesses.","title":"Tell HN: Avoid Cerebras if you are a founder","updated_at":"2026-04-26T10:19:21Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"alcasa"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"OpenAI is partnering with Cerebras to add 750MW of compute in 10B USD deal"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://openai.com/index/cerebras-partnership/"}},"_tags":["story","author_alcasa","story_46622763"],"author":"alcasa","children":[46622956,46624557,46629128,46630457,46631546,46642587,46707952],"created_at":"2026-01-14T20:32:49Z","created_at_i":1768422769,"num_comments":9,"objectID":"46622763","points":25,"story_id":46622763,"title":"OpenAI is partnering with Cerebras to add 750MW of compute in 10B USD deal","updated_at":"2026-03-05T23:24:13Z","url":"https://openai.com/index/cerebras-partnership/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"Tiberium"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Gemma 4 on Cerebras - The Fastest Inference Is Now Multimodal"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.cerebras.ai/blog/gemma-4-on-cerebras-the-fastest-inference-is-now-multimodal"}},"_tags":["story","author_Tiberium","story_48729020"],"author":"Tiberium","children":[48729083,48729086,48733021],"created_at":"2026-06-30T06:04:53Z","created_at_i":1782799493,"num_comments":8,"objectID":"48729020","points":24,"story_id":48729020,"title":"Gemma 4 on Cerebras - The Fastest Inference Is Now Multimodal","updated_at":"2026-07-01T19:48:17Z","url":"https://www.cerebras.ai/blog/gemma-4-on-cerebras-the-fastest-inference-is-now-multimodal"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"noob_hardy"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Post by ex-employee.
In the cutthroat AI world, Cerebras Systems, despite colossal wafers and bold claims, faces collapse. Its fragile foundation stems from brittle hardware, astonishingly narrow in practical application. Optimized primarily for LLaMA model finetuning, touted "speed boosts" often use speculative decoding, masking that many "supported" models run inefficiently, if at all. This rigidity mirrors a Formula 1 engine demanding one fuel under lab conditions.
This compounds with a poorly conceived software and kernel stack. The current infrastructure restricts Cerebras systems almost exclusively to LLaMA-family transformer LLMs. Modern vision pipelines, diffusion models, or other AI paradigms are beyond reach. In a rapidly evolving field, this narrow focus is a death sentence. A complete software overhaul is desperately needed, yet internal motivation is absent, hinting at leadership either oblivious or unwilling to invest in survival.\nTechnical woes are exacerbated by third-class ML expertise and rudderless, top-down leadership. Non-technical decision-makers lead to misguided strategies and poor adaptation. ML expertise is subpar and highly politicized; questionable ideas from figures like Head of AI Ganesh Venkatesh are reportedly prioritized, wasting resources.\nAn extreme resource shortage \u2013 Cerebras hardware and even GPUs \u2013 cripples meaningful work, making numbers in reports untrustworthy fiction. The chasm between rhetoric and reality widens.
A palpable resignation within the ranks further indicts the company. Employees are "waiting it out" or trading stocks, not contributing. This disengagement highlights deep problems, where personal finance trumps innovation.
Cerebras appears a bloated, incompetent, and culturally bankrupt company. Any stock price would be a scam. The rare "A players" have left or are leaving.
The writing is on the wall. Cerebras, despite grand pronouncements, is a company built on sand, destined to crumble under its limitations and poor choices.
Get out while you still can."},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras: The AI hardware giant facing imminent collapse"}},"_tags":["story","author_noob_hardy","story_45477162","ask_hn"],"author":"noob_hardy","children":[45478517,45479244,45480572,45486695,45510370],"created_at":"2025-10-04T22:08:08Z","created_at_i":1759615688,"num_comments":7,"objectID":"45477162","points":22,"story_id":45477162,"story_text":"Post by ex-employee.
In the cutthroat AI world, Cerebras Systems, despite colossal wafers and bold claims, faces collapse. Its fragile foundation stems from brittle hardware, astonishingly narrow in practical application. Optimized primarily for LLaMA model finetuning, touted "speed boosts" often use speculative decoding, masking that many "supported" models run inefficiently, if at all. This rigidity mirrors a Formula 1 engine demanding one fuel under lab conditions.
This compounds with a poorly conceived software and kernel stack. The current infrastructure restricts Cerebras systems almost exclusively to LLaMA-family transformer LLMs. Modern vision pipelines, diffusion models, or other AI paradigms are beyond reach. In a rapidly evolving field, this narrow focus is a death sentence. A complete software overhaul is desperately needed, yet internal motivation is absent, hinting at leadership either oblivious or unwilling to invest in survival.\nTechnical woes are exacerbated by third-class ML expertise and rudderless, top-down leadership. Non-technical decision-makers lead to misguided strategies and poor adaptation. ML expertise is subpar and highly politicized; questionable ideas from figures like Head of AI Ganesh Venkatesh are reportedly prioritized, wasting resources.\nAn extreme resource shortage \u2013 Cerebras hardware and even GPUs \u2013 cripples meaningful work, making numbers in reports untrustworthy fiction. The chasm between rhetoric and reality widens.
A palpable resignation within the ranks further indicts the company. Employees are "waiting it out" or trading stocks, not contributing. This disengagement highlights deep problems, where personal finance trumps innovation.
Cerebras appears a bloated, incompetent, and culturally bankrupt company. Any stock price would be a scam. The rare "A players" have left or are leaving.
The writing is on the wall. Cerebras, despite grand pronouncements, is a company built on sand, destined to crumble under its limitations and poor choices.
Get out while you still can.","title":"Cerebras: The AI hardware giant facing imminent collapse","updated_at":"2026-04-26T11:02:21Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"rntn"},"title":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"Cerebras's Condor Galaxy AI supercomputer takes flight carrying 36 exaFLOPS"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["cerebris"],"value":"https://www.theregister.com/2023/07/20/cerebras_condor_galaxy_supercomputer/"}},"_tags":["story","author_rntn","story_36800196"],"author":"rntn","children":[36800301,36800302,36800758,36801349],"created_at":"2023-07-20T13:33:09Z","created_at_i":1689859989,"num_comments":5,"objectID":"36800196","points":21,"story_id":36800196,"title":"Cerebras's Condor Galaxy AI supercomputer takes flight carrying 36 exaFLOPS","updated_at":"2024-09-20T14:32:25Z","url":"https://www.theregister.com/2023/07/20/cerebras_condor_galaxy_supercomputer/"}],"hitsPerPage":50,"nbHits":384,"nbPages":8,"page":0,"params":"query=Cerebris&tags=story&hitsPerPage=50&advancedSyntax=true&analyticsTags=backend","processingTimeMS":8,"processingTimingsMS":{"_request":{"roundTrip":14},"afterFetch":{"format":{"highlighting":1,"total":1},"merge":{"mergeLoop":{"prepareNextHit":1,"total":2},"total":2},"total":2},"fetch":{"query":4,"total":5},"total":8},"query":"Cerebris","serverTimeMS":11}