DeepSeek Watch posted an update
MiaAIlab says DeepSeek V4.1 Flash cost less than $3 to process more than 500 million tokens through its API, and reports peak speeds of roughly 315 tokens per second for decoding and 52,000 for prefill.
Why it mattersThat would be a striking serving datapoint for long-context and agent workloads, but the post gives no test setup or independent verification. Treat it as a result worth reproducing, not a settled performance record.
Discuss: What test setup would you want to see before accepting these cost and throughput figures as a useful benchmark?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.