Unverified50% confidenceBenchmarkExact time
The model's peak decoding speed on the test platform is approximately 274 tokens/second.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/8/2026
First Seen
Valid until: 10/22/2026
Sources
Related Entities
Related Claims
UnverifiedThe model gave the correct answer of 34 in a multimodal image recognition test counting pencils, at a speed of 55 tokens per second at that time.71% similarUnverifiedThe 26-billion-parameter Gemma (MoE) achieved a measured speed of nearly 31 tokens/second, about one-third faster than GPT-OSS 20B.70% similarUnverified2.4万亿参数模型在BF16精度下考虑KV Cache和激活值后实际约需1.4-1.5TB显存68% similarUnverifiedIn Path A, an RTX 3060 runs the front layers and an AMD AI Max+ 395 runs the back layers, improving Prefill speed from 136.69 token/s to 319.1 token/s under IQ3 quantization with 64K long input.67% similarUnverifiedAccording to the Chinchilla scaling law, the optimal training data volume for a 30B model is approximately 600B–1T tokens.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/987101API
curl https://kongchang.com/api/v1/knowledge/claims/987101MCP
get_claim(id=987101)