MiniMax launches M3.1 Flash model at half previous price
MarkTechPost: MiniMax Just Launched M3.1 Flash: 428B MoE Agentic Model at $0.15/M Input
MiniMax has just released MiniMax M3.1 Flash on its API, priced at $0.15 per 1M input tokens. MiniMax also confirmed that Space Bunny was an earlier M3.1 Flash build. That stealth model processed over 200T tokens. For developers, the message is near-frontier agentic scores at Flash-tier cost. Model weights are due in late October, 2026. TL;DR Size: 428B total parameters, about 23B active (MoE); 1,000,000-token context Runs on: MiniMax API, M Plan and MiniMax Code today. Self-hosting h
What happened
MiniMax released M3.1 Flash, a 428-billion-parameter mixture-of-experts model available through its API at $0.15 per million input tokens. The model processes text, images, and video inputs and scores 42 on the Artificial Analysis Intelligence Index, up 13 points from its predecessor M3. Model weights are scheduled to arrive in late October 2026.
What each outlet adds
- MarkTechPost ↗: MiniMax confirmed that Space Bunny was an earlier M3.1 Flash build that processed over 200 trillion tokens.
- The Decoder ↗: The model hits 73.7 percent on the DeepSWE v1.1 coding benchmark for under $0.30 per task and matches Deepseek V4.1 Flash on both performance and price, part of an ongoing competitive pricing battle between Chinese AI developers.
Why it matters
The release demonstrates near-frontier agentic performance at Flash-tier pricing costs. This reflects an ongoing competitive pricing race between Chinese AI developers, with neither this model nor its closest competitor winning across all performance measures.
Get smart on it
MiniMax released M3.1 Flash, a mixture-of-experts model with 428 billion total parameters and about 23 billion active parameters that can process text, images, and video inputs. The model is available through MiniMax's API at $0.15 per million input tokens and $0.60 per million output tokens, priced at half the rate of MiniMax's previous M3 model. M3.1 Flash performs near the top on agentic office work benchmarks such as Automation-Bench, though it scores lower than competing models on certain coding tasks. The model features always-on reasoning that developers cannot disable, with five effort levels available to choose from, and model weights are scheduled to arrive in late October 2026.
Relevant courses
Want to go deeper than the news? Explore live, cohort-based AI courses taught by practitioners.
Browse AI courses on MavenRelated news
Nvidia launches new open models to run agents in the enterprise and on PCs
Nvidia is leaning hard into the trend of enterprise companies wanting to run their own open models in their own data centers and also trying to make on-device, private AI agents a thing.
Kling AI drops version 4.0 with stereo audio, multilingual lip sync, and a lighter Flash variant
Kling AI has released version 4.0 of its video generator with support for clips up to 30 seconds, ten keyframe images as input, and prompt comprehension expanded to 8,000 tokens. In comparisons with Wan 3.0 and Seedance 2.5, Kling 4.0 holds its own but doesn't pull ahead, which could be an early sign that the breakneck pace of progress in AI video models is starting to level off. The article Kling AI drops version 4.0 with stereo audio, multilingual lip sync, and a lighter Flash variant
Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from Meta
Open-weight models let researchers and companies build AI systems without relying on proprietary providers, potentially accelerating innovation across industries.
Also reported by The Verge, The Decoder, TechCrunch and SiliconANGLE
Mistral AI Releases Timestral: A 209M Forecasting Model Family That Outperforms Google’s TimesFM-3
Mistral AI has released Timestral, its first family of time series foundation models. Timestral is a set of 4 zero-shot forecasting models from Mistral AI, sized from 5M to 209M parameters. The largest checkpoint ranks first among pretrained, zero-shot models on GIFT-Eval, TIME and fev-bench. It does this without a GPU at inference. For developers, this Timestral time series foundation model means probabilistic forecasts on standard server CPUs, with open weights and inference code. TL;DR
OpenAI will reportedly ship at least one model this year but its flagship still needs alignment work
OpenAI is racing to finish its next flagship model, which uses the same looped transformer technique as Astra to boost math and coding performance. The tradeoff is that looping makes safety monitoring harder. Microsoft accidentally revealed details about how the technique works on a public page. Whether or not the flagship is ready in time, employees say one or two new models will likely ship by year's end. The article OpenAI will reportedly ship at least one model this year but its fla
Open From Data Center to Desktop: NVIDIA Expands Nemotron Family With Highest Efficiency Super VL and Bolt Models
NVIDIA today released Nemotron 3.5 Super VL, an open 120-billion-parameter model that delivers the fastest task completion and highest accuracy for specialized always-on agents. Joining it are two new edge models: Nemotron 3.5 Bolt 74B-A9B 24GB, optimized to balance intelligence and speed per GB for local agentic workloads, and Nemotron 3.5 Bolt E4B 4GB, optimized for intelligence per GB for fast conversational agents. Super VL and Bolt 74B-A9B are available now, with Bolt E4B coming Oct. 23. T
Also reported by The Decoder, SiliconANGLE and MarkTechPost