🔍 Read the full analysis: MiniMax M3.1 Flash: Leaderboard Position Vs. Newsroom Test Performance on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
MiniMax released M3.1 Flash on October 19, 2026, at $0.15 per million input tokens and $0.60 per million output tokens. Artificial Analysis gave it an Intelligence Index score of 43 and measured 120 tokens per second, while MiniMax advertises speeds above 200 tokens per second. ThorstenMeyerAI.com also described a 24-article newsroom test, but the supplied material ends before reporting its scores, so no newsroom performance result can be verified here.
MiniMax released M3.1 Flash on October 19, pricing it at $0.15 per million input tokens and $0.60 per million output tokens. The launch presents the model as a faster, lower-cost option for agent work, but independent measurements reported by ThorstenMeyerAI.com put its output speed below the company’s claim, and the source material does not include the results of the publication’s separate newsroom writing test.
MiniMax says M3.1 Flash is available through its API, M Plan subscription, MiniMax Code and OpenRouter. The model has a stated one-million-token context window, adjustable reasoning effort and a cached-input price of $0.015 per million tokens. OpenRouter made it free for its first 72 hours, according to the report. Compared with M3, listed at $0.30 per million input tokens and $1.20 per million output tokens, the new model’s per-token prices are half as high.
The company says everyday output exceeds 200 tokens per second, with a median above 170, and describes that as nearly three times M3’s speed. Artificial Analysis, which evaluates models using a consistent set of ten tests, measured 120 tokens per second and 1.08 seconds to first token. Its Intelligence Index score for M3.1 Flash was 43, below the listed scores of 45 for DeepSeek V4.1 Flash 1017 and GLM-5.3, and 46 for Xiaomi’s MiMo-V2.6-Pro.
ThorstenMeyerAI.com says it tested the model on 24 articles derived from 12 news sources, producing English and German versions, with two judges scoring each article against its source. However, the supplied source text stops at the results table heading and contains no scores or findings. It therefore does not establish how M3.1 Flash performed on newsroom accuracy, factual invention, clarity or other writing criteria.
Infographic 01 · MiniMax M3.1 Flash
A step behind on the index. Ahead of every open model in our newsroom.
An independent intelligence index puts MiniMax M3.1 Flash a step behind the open-weight leaders. Writing real news from real sources, it beat all of them.
Independent index
Artificial Analysis Intelligence Index, 10 evaluations. Bar length on a 0–50 scale.
Artificial Analysis lists M3.1 Flash as proprietary until its weights are released.
Our newsroom test
Mean score, English articles, 1–5 by GPT-6.1 Sol and Claude Opus 5.5 with the source in context. Bar length on a 0–5 scale.
In German, M3.1 Flash (3.36) also leads every open model, but GPT-6 Luna is clearly ahead (3.62).
Sources: Artificial Analysis, MiniMax-M3.1-Flash and open-weights models (20 Oct 2026); Thorsten Meyer AI newsroom benchmark, 12 sources × English + German, run 20 Oct 2026.
Powered by Thorsten Meyer AI
Price Claims Meet Independent Tests
The release highlights a practical question for teams choosing models: whether a lower price per token also produces lower costs for a completed task. MiniMax claims M3.1 Flash costs about 70% less per agent task than M3, but that is a company figure, and the provided material does not supply the methodology behind it. Artificial Analysis found the model used 180 million tokens to complete its index, compared with a median of 99 million. That is a token-count observation for that evaluation, not a measure of newsroom costs or a universal estimate of usage.
The difference between MiniMax’s speed claim and the independent measurement also matters to developers planning response times. The two figures may reflect different test conditions; the source does not provide enough detail to reconcile them. The index score and speed measurement offer a comparison under Artificial Analysis’s evaluation, while the newsroom test could have addressed real-source writing quality. Its missing results prevent readers from judging that part of the report.
Products Worth Considering
Apple iPad Air 13-inch (M4), 128GB, Wi-Fi 7, Blue
Dell UltraSharp U4025QW 40 5K2K WUHD Curved LED Monitor (21:9)
Apple Mac mini Desktop Computer with M6 Chip (2026)
As an affiliate, we earn on qualifying purchases.
From Space Bunny to M3.1 Flash
Before the named release, the model was available anonymously on OpenRouter as Space Bunny Alpha. According to ThorstenMeyerAI.com, it appeared on September 23, led OpenRouter’s usage ranking for ten days and processed 56.7 trillion tokens on that platform before being removed on October 5. MiniMax says total usage across providers exceeded 200 trillion tokens. Those figures are attributed to the report and company, respectively.
The report says MiniMax revealed the model’s identity on October 19 and described it as an early version of M3.1 Flash. The company has promised to release the weights on October 23, but the supplied material does not say that release has occurred. Until then, Artificial Analysis lists the model as proprietary. MiniMax has not disclosed an official parameter count or license. A parameter estimate published by another outlet is excluded from the report because MiniMax has not confirmed it.
MiniMax’s own benchmark charts show mixed comparisons. Its model leads the listed table on Automation-Bench at 62.2, while DeepSeek V4.1 Flash 1017 scores higher on DeepSWE and Terminal-Bench. The report cautions that MiniMax selected the comparison models and that its results are self-reported; it says the company’s security results, including an 82% CyberGym score, have not been independently reproduced.
“More than 200 tokens per second in everyday use and a median above 170.”
— MiniMax
Newsroom Scores Are Missing
The largest evidence gap is the newsroom benchmark. The report describes its setup and says results were tabulated for English and German articles, but the supplied text ends before the figures and conclusions. It is not possible from this material to state whether M3.1 Flash performed well, poorly or comparably to other models in that test.
Other open questions include whether the promised weights will be released on October 23, under what license they might be distributed, and what the model’s parameter count is. The report also says MiniMax has not explained what happened to prompts submitted during the anonymous Space Bunny period. The provider could retain prompts during that period, although it was not allowed to train on them, according to the source.
The speed results cannot be treated as a direct contradiction without matching test conditions. MiniMax’s everyday-use claim and Artificial Analysis’s evaluation may use different hardware, settings or workloads; the source does not specify enough to compare them precisely. MiniMax’s benchmark results also remain company-reported rather than independently reproduced in the material provided.
Watch for Weights and Test Data
The next stated milestone is MiniMax’s planned October 23 weight release. Its availability, license and technical details should be checked against the company’s actual release rather than treated as settled. A public weight release could allow outside developers and evaluators to test the model beyond the hosted services listed at launch.
For the newsroom comparison, the missing English and German scores are needed before drawing a conclusion about article quality or source fidelity. Readers should also look for the full benchmark methodology, including the scores assigned by each judge and comparisons with other models run under the same prompt and conditions. Until that information is available, the supported conclusion is limited: M3.1 Flash has a low listed token price, an index score of 43, and a measured speed below MiniMax’s advertised figure in Artificial Analysis’s evaluation.
Key Questions
When did MiniMax release M3.1 Flash?
MiniMax revealed and released M3.1 Flash on October 19, 2026, according to ThorstenMeyerAI.com.
How much does M3.1 Flash cost?
The listed price is $0.15 per million input tokens and $0.60 per million output tokens. Cached input is listed at $0.015 per million tokens.
How fast is M3.1 Flash?
MiniMax claims everyday speed above 200 tokens per second, with a median above 170. Artificial Analysis measured 120 tokens per second and 1.08 seconds to first token in its evaluation.
How did it perform in the newsroom test?
The source describes a test of 24 articles in English and German, scored by two judges, but the supplied material does not include the results. Its newsroom performance cannot be determined from the available information.
Are the model weights available?
At the time covered by the source, the weights had not been released. MiniMax said they would arrive on October 23; the source does not confirm a later release or specify a license.
Source: ThorstenMeyerAI.com
