Unverified50% confidenceDecision RuleExact time
MiniMax M3.1 Flash's thinking tokens count toward the max_tokens limit, so reusing the 4K or 8K caps from the M2.x era may cause truncation with finish_reason: "length"; MiniMax recommends raising max_tokens to 131,072, with an upper limit of 524,288.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Related Entities
Related Claims
UnverifiedMiniMax M3.1 Flash defaults to the max setting when the effort field is not passed in the request, and the model always performs thinking.78% similarUnverifiedMaximum output tokens: MiniMax M3.1 Flash is 512K, DeepSeek V4.1 Flash 1017 is 384K, and Claude Haiku 5.5 is 128K.77% similarUnverifiedBased on public data for high power plus MLX or Ollama, the decoding speed of different models on M4 Max is roughly a ladder as follows: Nemotron 3 Nano about 160 tokens/s, Qwen 3.5 35B at 107 to 130 tokens/s, Gemma 4 26B about 75 tokens/s, DeepSeek V4 Flash about 27 tokens/s, and dense 27B about 22 tokens/s.72% similarUnverifiedThe MiniMax H3 standard workflow requires a minimum of 16GB VRAM to run, while 6GB graphics cards can also participate in partial workflows through optimizations such as FP8, INT4 quantization, and chunked inference Tiling.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1067178API
curl https://kongchang.com/api/v1/knowledge/claims/1067178MCP
get_claim(id=1067178)