AIHubMix

MiniMax M3.1 Flash is available on AIHubMix for token-billed calls under the model ID `minimax-m3.1-flash`, with input priced at $0.165 per million tokens and output at $0.66 per million tokens.

Calculated per thousand repair tasks, Claude Haiku 5.5 costs about $0.70, DeepSeek V4.1 Flash 1017 about $1.35, and MiniMax M3.1 Flash about $4.45 by default and about $0.64 at the low setting.

In the dual-bug Python function repair benchmark (215 hidden tests), omitting effort (max) consumes an average of 6,364 reasoning tokens, takes 56 seconds, and costs $0.0045 per run; effort=low consumes an average of 591 tokens, takes 9.5 seconds, and costs $0.0006 per run.

In AIHubMix testing, every response carries "model": "deepseek-v4-1-flash-261017", so unannounced model swaps can be detected by storing the model field in responses and re-running the regression suite.

If the same session stops after the 21st call (the prompt has not crossed 100K), Haiku 5.5 costs $0.040, and DeepSeek V4.1 Flash 1017 costs $0.077 at peak and $0.039 off-peak.

Under the Messages protocol, DeepSeek V4.1 Flash 1017 requests without an effort level set do not return a thinking block, and the effort level must be set explicitly.

DeepSeek V4.1 Flash 1017's batch and night-time agent tasks scheduled outside weekday UTC 01:00–04:00 and 06:00–10:00 get half-price billing; AIHubMix offers a limited-time 30% discount on 1017.

In a 40-call session, 19 of Haiku 5.5's requests cross the 100K token line, and these 19 requests contribute $0.248, accounting for the vast majority of the session's $0.288.

API calls based on AIHubMix require AIHUBMIX_API_KEY, the base_url is https://aihubmix.com/v1, and the model IDs are deepseek-v4.1-flash-1017 and deepseek-v4-pro-0813 respectively.

Third-party platform AIHubMix list prices: Flash 1017 is $0.3098 input / $1.2392 output, and Pro 0813 is $1.3944 input / $4.1832 output.

Under an assumed Agent bug-fixing session of 40 calls, 1.2 million input tokens (90% cache hit rate) and 60,000 output tokens, calculated at AIHubMix list prices, Flash 1017 costs $0.118 at peak hours and $0.059 off-peak; Pro 0813 costs $0.469 at peak and $0.233 off-peak; Pro costs about four times as much per session as Flash.

On AIHubMix, MiniMax M3.1 Flash's input price is about 44% lower than M2.7's ($0.165 vs. $0.2958) and about 43% lower than M3's ($0.165 vs. $0.288), with the same ratios on the output side.

At AIHubMix pricing (input $0.165, output $0.66, cache reads $0.0165 per 1M tokens), the default max setting makes thinking tokens account for 95% of output (6,364/6,673), with thinking cost more than 10 times that of the low-intensity setting and latency close to 6 times.

On the same two-bug repair task, omitting the effort field (default max) consumes an average of 6,364 thinking tokens, 56 seconds, and $0.0045 per repair, whereas the low setting uses an average of only 591 thinking tokens, 9.5 seconds, and $0.0006.

On AIHubMix's Gemini endpoint, tool_choice="auto" must be passed together when using tools, otherwise function calling will fail.

An implicit cache hit requires the content, model, and parameters to all be identical, so changing the thinking level mid-conversation breaks cache hits.

AIHubMix testing on October 19, 2026 shows that the model field of all responses reads deepseek-v4-1-flash-261017, and that all three reasoning_effort values low/high/max are accepted and return HTTP 200.

On October 19, MiniMax M2.5/M2.7 were closed to new users and plan requests were routed to M3.1 Flash; on the same day `minimax-m3.1-flash` launched on AIHubMix on a pay-per-token basis.

DeepSeek V4.1 Flash was released on September 10, with the 1017 update launched on October 17; Claude Haiku 5.5 was released on October 7; MiniMax M3.1 Flash was named on September 27 and opened for per-token API calls on AIHubMix on October 19.