🔍 Read the full analysis: MiniMax M3.1 Flash: Comparing Its Leaderboard Standing With Our Test Results on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
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 scored it 43 on its Intelligence Index, below several open-weight models, while the publisher’s 24-article newsroom test is described but its results are missing from the supplied material.
MiniMax released M3.1 Flash on October 19, pricing the model at $0.15 per million input tokens and $0.60 per million output tokens, as it targets faster, lower-cost agent work. An independent Artificial Analysis evaluation gave it an Intelligence Index score of 43, behind several open-weight competitors; results from a separate 24-article newsroom test are not included in the source material provided for this report.
MiniMax says M3.1 Flash costs half as much per token as its predecessor, M3, and about 70% less per agent task. Those savings are company claims, not independently verified cost comparisons in the available material. The model is available through the MiniMax API, M Plan subscription, MiniMax Code and OpenRouter. OpenRouter offered it free for its first 72 hours. Listed prices are $0.15 per million input tokens, $0.60 per million output tokens and $0.015 per million cached input tokens.
MiniMax claims everyday output speeds above 200 tokens per second, with a median above 170, and says the model is “nearly three times faster” than M3. Artificial Analysis measured 120 tokens per second and 1.08 seconds to first token in its evaluation. The index places M3.1 Flash at 43, compared with 45 for DeepSeek V4.1 Flash 1017 and GLM-5.3, and 46 for Xiaomi’s MiMo-V2.6-Pro. The different speeds reflect separate measurements; the source does not provide enough methodological detail to reconcile them.
MiniMax’s own benchmark chart shows M3.1 Flash leading on Automation-Bench with 62.2, while DeepSeek V4.1 Flash leads on DeepSWE and Terminal-Bench, at 77.0 and 42.3 respectively. MiniMax selected the competitors and published the figures itself. The source says these results, including the company’s security evaluation, have not been independently reproduced. It also notes that the model used 180 million tokens to complete Artificial Analysis’s index, against a reported median of 99 million; token use can affect total costs even when token prices are low.
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 Meets Real-World Performance
M3.1 Flash’s appeal rests on a combination of low listed prices, claimed speed and suitability for agent tasks. The independent index offers a more qualified picture: it places the model below several open-weight rivals and records lower throughput than MiniMax’s stated everyday figure. For teams choosing a model, the difference between per-token price and the total tokens needed to complete a task may matter as much as the headline rate.
The newsroom comparison could address a distinct practical question: whether the model can turn source material into accurate, useful reporting without adding unsupported details. The publisher says it tested 24 articles in English and German, using two judges and the same production prompt used for other models. But the supplied source cuts off before the scores and comparisons appear. Without those results, readers cannot assess the test’s findings or weigh them against the leaderboard.
As an affiliate, we earn on qualifying purchases.
From Space Bunny to M3.1
Before the public release, the model ran anonymously on OpenRouter under the name Space Bunny Alpha. The source says it went live on September 23, led OpenRouter’s usage ranking for 10 days and processed 56.7 trillion tokens there before being removed on October 5. MiniMax says total usage across services exceeded 200 trillion tokens. These are figures attributed to the source’s account and the company, rather than independent measures of model quality.
MiniMax says M3.1 Flash has a one-million-token context window, adjustable reasoning effort and “more aggressive sparsity” than its predecessor. It has not yet released the weights or announced a license or official parameter count. The source says MiniMax scheduled the weights for October 23. Until they are available, Artificial Analysis lists the model as proprietary, and outsiders cannot inspect or run the weights independently.
MiniMax’s benchmark comparisons also have limits. Its chart includes selected rivals, and some entries are marked “not reported.” The source flags a claim that M3.1 Flash ranks first among Chinese models on Agents’ Last Exam: DeepSeek’s model card reports a higher score for DeepSeek V4.1 Flash 1017, though the setups may differ, and MiniMax’s chart does not include DeepSeek in that row. The available evidence does not settle the comparison.
““nearly three times faster” than M3”
— MiniMax
Missing Test Scores and Release Details
The source material does not include the newsroom test’s score table or its conclusions. It describes the test setup—12 sources, each written in English and German, and reviewed by GPT-6.1 Sol and Claude Opus 5.5—but stops before presenting results. No claim about M3.1 Flash’s relative newsroom performance can be made from the material provided.
Other open questions include whether MiniMax will release the weights on October 23 as scheduled, what license will apply, and what the model’s parameter count is. The source mentions an outside report of 428 billion total and 23 billion active parameters, but says MiniMax has not confirmed those figures. It also says MiniMax has not explained what happened to prompts retained during the anonymous Space Bunny period. The provider was permitted to keep prompts, though not to train on them, according to the source.
MiniMax’s benchmark and security figures remain company-reported in the supplied account, while independent evaluations measure different aspects of performance. The source provides no independent reproduction of the company’s benchmark results and no basis for treating its cost-per-task estimate as a verified outcome across users’ workloads.
Watch for Weights and Test Data
The next stated milestone is the planned October 23 weights release. Its timing, license and parameter details will show whether external users can examine and deploy the model beyond the hosted services already listed. Any release should also make it possible for independent evaluators to run reproducible tests, although the source does not say when such evaluations will appear.
For the newsroom comparison, the missing score table and model-by-model results are needed before the test can support a performance judgment. Readers should also look for further detail on the judges’ scores, the free OpenRouter endpoint and the limits used in testing. Until those results are published, the evidence supports a narrow conclusion: M3.1 Flash is a newly released, low-priced model with a solid but not leading score on the cited independent index, while its performance on the described newsroom test remains unknown.
Key Questions
When did MiniMax release M3.1 Flash?
MiniMax released M3.1 Flash on October 19, 2026, according to the supplied report.
How much does M3.1 Flash cost?
The listed rates are $0.15 per million input tokens, $0.60 per million output tokens and $0.015 per million cached input tokens. MiniMax also claims about 70% lower cost per agent task than M3, but that comparison is a company claim.
How did M3.1 Flash score on Artificial Analysis?
Artificial Analysis gave it an Intelligence Index score of 43. The supplied comparison lists DeepSeek V4.1 Flash 1017 and GLM-5.3 at 45, and MiMo-V2.6-Pro at 46.
Are the newsroom test results available?
The source describes a test of 24 articles in two languages, assessed by two judges, but the provided material ends before the scores. Its findings cannot be established from the text supplied.
When are the model weights expected?
MiniMax said the weights would arrive on October 23. The source does not confirm that they have been released or state what license will apply.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
