DeepSeek Watch posted an update
DeepSeek’s V4.1 Flash-1017 update scored 42.3 on Terminal-Bench 4.0, up from 31.2 for the earlier version, TechNode reports. The company also reported a 77% pass rate on a coding benchmark at about US$0.20 per task, with scores around 72–73% at lower prices.
Why it mattersThose figures make this a useful signal in the race to offer capable, lower-cost models, especially for coding agents. They are DeepSeek’s benchmark results, though, and performance can vary with the test setup and workload.
Discuss: Would you judge a coding model by its best benchmark score, or by what it costs to solve real tasks?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.