Unverified50% confidenceFactExact time
Renting an RTX 5090 on Vast.ai ($0.584 per hour), using bf16 and torch.compile, Keera's pretraining throughput was about 234,000 tokens/second, the 2B-token training took 2 hours 23 minutes, and the validation loss was 1.501.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/18/2026
First Seen
Valid until: 11/1/2026
Sources
Related Entities
Related Claims
UnverifiedThe training cost only $164 in rented GPU time, running for 54.6 hours on a rented H100.68% similarUnverifiedThe vendor claims a median Token count of 2700 per ticket, 1.2 API calls, an inference cost of $0.003, and a wall-clock time of 3.5 seconds.62% similarUnverifiedEstimated at 70 tokens/second, approximately 252 million tokens can be generated in one hour, translating to an electricity cost of approximately €0.48 per million tokens.61% similarUnverified云端 AI 方案网络延迟往返时间通常在 200ms-2s，GPT-4 级别模型费用约 $30/百万 token；端侧方案本地推理可在 50-200ms 内完成首 token 生成且零边际成本60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1057716API
curl https://kongchang.com/api/v1/knowledge/claims/1057716MCP
get_claim(id=1057716)