🔍 Read the full analysis: MiniMax M3.1 Flash: Does Its Newsroom Test Match Its Leaderboard Rank? on ThorstenMeyerAI.com
Get privacy and security gear 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 measured an Intelligence Index score of 43 and 120 tokens per second, while a newsroom test covered 24 articles; the source material provided here does not include that test’s results table, so its performance cannot be independently summarized.
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. A review published the next day compared MiniMax’s claims with Artificial Analysis measurements and a newsroom test of 24 articles, but the source material available for this report does not include the newsroom results table, leaving its central editorial-performance comparison unresolved.
MiniMax describes M3.1 Flash as a low-cost, fast model for agent work. The company says it is nearly three times faster than M3 in everyday use, with speeds above 200 tokens per second and a median above 170. It also claims the new model costs about 70% less per agent task than its predecessor. Those are company claims, not independent measurements. The model has a stated one-million-token context window, adjustable reasoning effort and a cached-input price of $0.015 per million tokens.
Artificial Analysis gives M3.1 Flash an Intelligence Index score of 43. In the comparison cited by the review, that trails MiMo-V2.6-Pro at 46, DeepSeek V4.1 Flash 1017 and GLM-5.3 at 45, and Step 5 Preview and Kimi K3 at 44. Artificial Analysis measured output speed at 120 tokens per second and time to first token at 1.08 seconds—below MiniMax’s stated everyday-use speed and below DeepSeek’s reported 211 tokens per second.
The newsroom evaluation used 12 real news sources in English and German, producing 24 articles. Two judges, GPT-6.1 Sol and Claude Opus 5.5, were asked to score source fidelity, invented facts, clarity, usefulness, structure and polish on a five-point scale. The supplied source describes the setup but cuts off before the scores and comparison table. It therefore does not establish whether M3.1 Flash performed well or poorly against other models on this test.
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 Versus Measured Performance
M3.1 Flash’s low token price may appeal to developers building high-volume agent workflows, where API costs can add up quickly. But price per token is only one part of operating cost. Artificial Analysis found the model used 180 million tokens to complete its index evaluation, compared with a median of 99 million among the models in that comparison. That is a raw token-use observation for the evaluation, not a general measure of every task; it suggests that output length can affect the cost advantage, depending on workload.
The separate newsroom test addresses a different question: whether the model can produce accurate, usable reporting from source material. That matters to publishers and other teams using models to transform documents into content, because fluency alone does not show whether a system preserves facts or avoids invention. Without the results table, readers cannot judge how M3.1 Flash performed on those criteria or compare it with the other models tested.
For now, the evidence points to a model with a competitive price and a solid independent index score, but not the leading score in the cited open-weight comparison. Its claimed speed and task-cost advantages also need to be weighed against third-party measurements and the token demands of a specific workload.
encrypted USB flash drives for secure file storage
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Space Bunny to M3.1
Before its public release, the model appeared anonymously on OpenRouter as Space Bunny Alpha. The review says it was available from September 23 until October 5, led OpenRouter’s usage ranking for 10 days and processed 56.7 trillion tokens on the platform. MiniMax separately reported more than 200 trillion tokens overall. Those figures describe different stated scopes and should not be treated as interchangeable.
MiniMax announced M3.1 Flash on October 19 and made it available through its API, M Plan subscription, MiniMax Code and OpenRouter. OpenRouter offered free access for the first 72 hours, according to the review. The company said the model’s weights would arrive on October 23, after the review’s October 20 publication date. Until release, Artificial Analysis listed it as proprietary. MiniMax had not published an official parameter count or a license in the cited material.
MiniMax’s own benchmark charts report mixed results against selected competitors. M3.1 Flash scored 62.2 on Automation-Bench, ahead of DeepSeek V4.1 Flash 1017 at 60.5, but DeepSeek scored higher on DeepSWE and Terminal-Bench. The review notes that the comparisons rely on company-reported results and that MiniMax selected the rivals. It also says MiniMax’s security figures came from an internal evaluation and had not been independently reproduced at the time.
“The model is “nearly three times faster” than M3.”
— MiniMax
AI language model performance testing tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Missing Newsroom Scores
The supplied review text does not include the newsroom benchmark’s score table, the tested comparison models, or its conclusions. That means it is not possible to report M3.1 Flash’s newsroom score, identify a winner, or say how often judges found invented details. The test methodology is described, but methodology alone cannot establish the outcome.
Other open questions include whether the promised weights arrived on October 23 and under what license, since the available account predates that date. MiniMax also had not confirmed a parameter count. The review notes a separate outlet’s estimate of 428 billion total parameters and 23 billion active parameters, but MiniMax had not confirmed those figures, so they should not be treated as established specifications.
The account also says prompts submitted during the Space Bunny Alpha period could be retained by the provider, although they would not be used for training. MiniMax had not explained what happened to that retained data in the cited material. The model’s handling of those earlier prompts and the independent reproducibility of its company-published benchmarks remain unresolved there.
As an affiliate, we earn on qualifying purchases.
Weights and Benchmark Results
The next stated milestone was MiniMax’s planned October 23 release of the model weights. If released, the weights and their license would clarify whether developers can run the model outside MiniMax’s hosted services and under what terms. The available source does not confirm that the release subsequently occurred.
For the newsroom comparison, the missing results are necessary to answer the question posed by the review: whether performance on a real editorial workflow matches the model’s standing on an independent index. A complete account would need to report scores by language and criterion, name the comparison models, and explain how judges handled disagreements. Until those details are available, the test’s outcome remains unknown.
Further independent evaluations could also test the company’s speed and agent-cost claims across defined tasks. For readers considering the model, the evidence available in the October 20 review supports comparing it on the intended workload rather than relying on a leaderboard position or a per-token price alone.
As an affiliate, we earn on qualifying purchases.
Key Questions
When was MiniMax M3.1 Flash released?
MiniMax released M3.1 Flash on October 19, 2026, according to the review dated October 20.
What 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 did M3.1 Flash perform on the newsroom test?
The available source describes a 24-article test but does not include its results table. Its scores and comparative performance therefore cannot be confirmed from the supplied material.
How does its independent index score compare with competitors?
Artificial Analysis assigned M3.1 Flash an Intelligence Index score of 43. The review lists several open-weight models scoring between 44 and 46 in the same comparison.
Were the model weights available at the time of the review?
No. The October 20 review said MiniMax planned to release the weights on October 23. The source material does not confirm whether that later release took place.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
