🔍 Read the full analysis: Why MiniMax M3.1 Flash Stands Out In Our Test Despite Its Leaderboard Position on ThorstenMeyerAI.com
Get hardware and tech essentials 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, promoting it as a fast, low-cost model for agent tasks. In a newsroom test of 24 articles, it stood out as a candidate for real-world writing evaluation, while independent index results placed it below several open-weight competitors. The source material does not provide the newsroom scores, so the size of its advantage cannot 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, as a newsroom test assessed how well it could turn real sources into articles. The test covered 24 English- and German-language articles, but the available source material does not include its scores, limiting what can be confirmed about how the model compared with alternatives.
MiniMax describes M3.1 Flash as a fast, low-cost model for agent work. It is available through the MiniMax API, M Plan subscription, MiniMax Code and OpenRouter. OpenRouter offered access without charge for the first 72 hours, according to the source report. The listed price is half the per-token rate for its predecessor, M3, and MiniMax says agent tasks cost about 70% less. That task-cost comparison is the company’s claim, not an independently established result.
The report evaluates the model using three kinds of evidence: MiniMax’s own benchmarks, Artificial Analysis’s independent index and a newsroom test using real news sources. The newsroom evaluation drew on 12 sources, ranging from a short press release to a longer report. Each was used to produce an English article and a German article, for 24 outputs in total. Two judges, GPT-6.1 Sol and Claude Opus 5.5, scored each article from 1 to 5 on source fidelity, invented facts, clarity, usefulness, structure and polish.
The supplied account describes that setup but cuts off before giving the newsroom results table. It therefore does not establish M3.1 Flash’s score, its ranking against other tested models or the size of any performance gap. The report says the model ran through OpenRouter’s free endpoint with a 32,000-token output limit. Those details are useful for understanding the test, but the missing results prevent a full comparison.
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
Why Article-Level Testing Matters
The comparison highlights why a model’s position on a broad leaderboard may not settle whether it is suitable for a particular job. Artificial Analysis gave M3.1 Flash an Intelligence Index score of 43, below several listed open-weight models, including MiMo-V2.6-Pro at 46 and DeepSeek V4.1 Flash 1017 and GLM-5.3 at 45. A newsroom test asks a narrower question: whether a model can produce a useful article from a particular source while avoiding unsupported details.
That distinction matters to publishers and other teams weighing quality, speed and cost. A low price per token does not automatically mean a cheaper completed task if the model uses more tokens; Artificial Analysis recorded 180 million tokens for its index evaluation, compared with a 99 million median. That is a measurement from that evaluation, not proof of how much any newsroom assignment will cost. Without the article scores and a like-for-like cost breakdown, the source does not show whether M3.1 Flash’s price offsets its token use in production.
Products Worth Considering
Logitech MX Keys S Wireless Keyboard, Graphite
Apple MacBook Pro 16.2-inch with M5 Pro, 18-Core CPU, 20-Core GPU, 48GB Memory, 1TB SSD, Wi-Fi 7, Silver
Ledger Flex Crypto Wallet
As an affiliate, we earn on qualifying purchases.
From Space Bunny to M3.1 Flash
Before its public release, the model appeared anonymously on OpenRouter under the name Space Bunny Alpha. The source report says it went live on September 23, led the platform’s usage ranking for 10 days and processed 56.7 trillion tokens there before being removed on October 5. MiniMax says it processed more than 200 trillion tokens overall; that figure is the company’s account.
MiniMax says M3.1 Flash has a one-million-token context window and adjustable reasoning effort. Its explanation of the design is limited to “more aggressive sparsity.” The company claims everyday output above 200 tokens per second and a median above 170, nearly three times faster than M3. Artificial Analysis measured 120 tokens per second and 1.08 seconds to the first token in its own evaluation, below the company’s speed claim. Differences in test conditions may affect the comparison.
MiniMax’s published benchmark chart, as transcribed in the source report, shows M3.1 Flash ahead on Automation-Bench with 62.2, compared with 60.5 for DeepSeek V4.1 Flash 1017. DeepSeek scored higher on DeepSWE and Terminal-Bench. MiniMax selected the models in its comparison, and not every rival appears in every row. The report also says the scores and security results were self-reported and had not yet been reproduced independently.
““more aggressive sparsity””
— MiniMax
Scores and Data Still Missing
The supplied source ends after introducing a results table for the newsroom test; it does not show the scores. That leaves the central claim in the headline—that M3.1 Flash stands out in the test despite its leaderboard position—unquantified in the material provided. The test design is described, but its outcome cannot be independently checked from this account.
The model weights were still pending, with MiniMax promising them for October 23. The source says there was no published license or official parameter count. It mentions an external report estimating 428 billion total parameters and 23 billion active parameters, but says MiniMax had not confirmed those figures. It also says MiniMax had not explained what happened to prompts retained during Space Bunny Alpha’s anonymous OpenRouter period. The provider was allowed to retain prompts but not train on them, according to the report.
Other open questions include whether independent evaluators will reproduce MiniMax’s benchmark and security results, and how the model’s speed and token use compare under matched tasks. The report’s comparison also notes that some rivals selected by MiniMax were older releases, while not all models had results listed for each benchmark.
Weights and Test Results Ahead
The next stated milestone is MiniMax’s promised October 23 weights release. Their publication would allow reviewers to inspect the model and assess it outside the company’s hosted services, subject to the license MiniMax provides. The available report does not confirm that the release has occurred.
A fuller judgment of M3.1 Flash’s newsroom performance requires the missing article scores, the competing models’ results and enough information to reproduce the test. Readers should also watch for independent checks of the company’s speed, cost and benchmark claims, along with clarification about data retained during the anonymous OpenRouter run. Until then, the evidence supports a comparison of the model’s stated price and published evaluations, but not a definitive claim about its newsroom-test ranking.
Key Questions
What is MiniMax M3.1 Flash?
It is a MiniMax language model released October 19, 2026. MiniMax markets it for agent tasks and lists prices of $0.15 per million input tokens and $0.60 per million output tokens.
Did M3.1 Flash win the newsroom test?
The source describes a 24-article evaluation but does not include the results table. Its score or rank cannot be confirmed from the supplied material.
How did it score on the independent index?
Artificial Analysis gave M3.1 Flash an Intelligence Index score of 43. Several listed open-weight models scored higher, including MiMo-V2.6-Pro at 46 and DeepSeek V4.1 Flash 1017 and GLM-5.3 at 45.
Are the model weights available?
At the time of the October 20 report, the weights had not been released. MiniMax said they would arrive October 23; the supplied material does not verify whether that happened.
How fast is M3.1 Flash?
MiniMax claims everyday speed above 200 tokens per second, while Artificial Analysis measured 120 tokens per second in its evaluation. The results come from different sources, and the report does not establish that their test conditions were identical.
Source: ThorstenMeyerAI.com
