🔍 Read the full analysis: The Newsroom Performance Of MiniMax M3.1 Flash Compared With Open Models on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- 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, pricing it at $0.15 per million input tokens and $0.60 per million output tokens. An independent index gave it a score of 43, below several open-weight rivals, while a newsroom test of 24 articles was described but its results were not included in the source material provided.
The model is available through the MiniMax API, M Plan subscription, MiniMax Code and OpenRouter. OpenRouter offered it free for its first 72 hours. The listed rate for cached input is $0.015 per million tokens. MiniMax says M3.1 Flash costs half as much per token as its M3 predecessor and about 70% less per agent task; those savings are company claims, not results independently verified in the review.
MiniMax describes the model as designed for fast, low-cost agent work, with a one-million-token context window and adjustable reasoning effort. It 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 a 1.08-second time to first token. The independent measurement is lower than the company’s stated speed, though the figures may reflect different testing conditions.
The review also describes a newsroom evaluation using 12 real news sources in English and German, producing 24 articles. Two judges scored source fidelity, invented details, clarity, usefulness, structure and polish. The source text supplied for this article ends before the results table, so it does not establish how M3.1 Flash scored or how it compared with other tested models.
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 Independent Performance
The launch adds a low-priced option for developers building agents and other high-volume applications, where token costs can accumulate. But price per token is only one part of operating cost: Artificial Analysis found M3.1 Flash used 180 million tokens to complete its index, compared with a median of 99 million. That result suggests its output volume may offset some savings, although it comes from a particular evaluation and is not a universal measure of cost.
On Artificial Analysis’ Intelligence Index, M3.1 Flash scored 43. The listed open-weight models MiMo-V2.6-Pro, DeepSeek V4.1 Flash and GLM-5.3 scored 46, 45 and 45, respectively. The index therefore places M3.1 Flash behind those models on its measured capabilities, even as the new model offers a lower price than many alternatives. Buyers weighing it for newsroom or business use will also need task-specific accuracy results, which cannot be determined from the incomplete newsroom findings provided.
As an affiliate, we earn on qualifying purchases.
From Space Bunny to M3.1 Flash
Before the release, the model ran anonymously on OpenRouter as Space Bunny Alpha. The review says it appeared on September 23, led the platform’s usage ranking for 10 days and processed 56.7 trillion tokens there before being withdrawn on October 5. MiniMax says it processed more than 200 trillion tokens overall; that figure is attributed to the company.
MiniMax’s benchmark charts compare M3.1 Flash with several rivals, including DeepSeek V4.1 Flash 1017, Kimi K3, GLM-5.3, GPT-5.6 Sol and Claude Opus 5. The review notes that some named competitors were older versions than the latest models available at the time of its report. In those company-reported results, DeepSeek leads M3.1 Flash on DeepSWE and Terminal-Bench, while M3.1 Flash leads on Automation-Bench. These are vendor-selected, self-reported comparisons, and the review says they had not been independently reproduced.
Although MiniMax calls the model part of an open-weight field, its weights were not yet available at release. Artificial Analysis listed it as proprietary for that reason. MiniMax said the weights would arrive October 23, but the material provided gives no confirmation that they were subsequently published.
“The company said M3.1 Flash delivers everyday speeds of more than 200 tokens per second and a median above 170.”
— MiniMax
As an affiliate, we earn on qualifying purchases.
Missing Scores and Release Details
The supplied review material does not include the newsroom test’s result table. It therefore supports the test design and judging criteria, but not a conclusion about M3.1 Flash’s newsroom performance or its ranking against the other models tested. The article should not infer those results from the setup alone.
Other details remain unconfirmed. MiniMax had not published the weights on the October 19 release date, and the review reports no announced license or official parameter count. A separate outlet reportedly put the model at 428 billion total and 23 billion active parameters, but MiniMax had not confirmed those figures. The company also had not explained what happened to prompts submitted during the anonymous Space Bunny period; the review says the provider could retain prompts but could not use them for training.
Benchmark comparisons also have limits. MiniMax selected its competitors and reported the scores in charts; some results were not listed for every model. The review flags a claim that M3.1 Flash ranked first among Chinese models on Agents’ Last Exam, while DeepSeek’s model card reported a different score for its own model, potentially using another setup. The available information does not resolve that discrepancy.
AI model performance testing tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Weights, Licensing and Full Test Results
The next stated milestone was MiniMax’s planned October 23 weights release. If published, the weights and accompanying license would clarify whether developers can run or adapt the model outside MiniMax’s hosted services. The supplied material does not confirm that release or specify its terms.
For a fuller performance assessment, readers would need the omitted newsroom scores, including comparisons across both languages and the two judges’ assessments. Independent replication of MiniMax’s benchmark claims, and more information about the conditions behind its speed figures, would also help establish how the model performs across tasks. Until those details are available, the clearest confirmed picture is a low launch price, mixed benchmark evidence and unresolved release information.
newsroom AI content generation tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
When did MiniMax release M3.1 Flash?
MiniMax released M3.1 Flash on October 19, 2026, according to the review. The company said its weights would follow on October 23, but the supplied material does not confirm that they were released.
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. MiniMax says the model costs about 70% less per agent task than M3, but the review presents that as a company claim.
Does M3.1 Flash outperform open-weight models?
Not across the evidence provided. Artificial Analysis scored it 43 on its Intelligence Index, below several listed open-weight models. MiniMax’s own benchmark charts show mixed results, and the newsroom test scores are missing from the supplied material.
What did the newsroom test measure?
The review says it tested 24 articles: 12 real news sources written once in English and once in German. Two judges scored source fidelity, invented facts, clarity, usefulness, structure and polish. The actual scores are not present in the provided text.
Were M3.1 Flash’s weights open at launch?
No. The review says the weights were not available on October 19; MiniMax had promised them for October 23. It also says no license or official parameter count had been published at the time covered.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
