Podcast
Oct 18, 2026 · 5 min · 8 segments
The October 17 release replaces the model behind deepseek-flash. DeepSeek reports stronger coding results and shorter answers. This edition covers 5 evidence-backed stories. AI-generated narration…
The company also released the updated weights under the MIT license.
The model card reports a score of 42.3 on terminal bench 4.0, up from 31.2 for the previous version.
DeepSeq says the model completes tasks faster and produces shorter answers.
The company also opened its search and page fetching APIs.
DeepSeq supplied the performance and cost claims.
HotTea has not reproduced them.
Terminal bench results can also vary with the benchmark version and the software used to run the test.
Run the new version against saved tasks.
Compare completed work, retries and total cost.
Then watch for independent tests of this revision.
The company also released the updated weights under the MIT license.
The model card reports a score of 42.3 on terminal bench 4.0, up from 31.2 for the previous version.
DeepSeq says the model completes tasks faster and produces shorter answers.
The company also opened its search and page fetching APIs.
DeepSeq supplied the performance and cost claims.
HotTea has not reproduced them.
Terminal bench results can also vary with the benchmark version and the software used to run the test.
Run the new version against saved tasks.
Compare completed work, retries and total cost.
Then watch for independent tests of this revision.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
