# SiliconANGLE — MiniMax launches M3.1 Flash model, says it cuts AI agent costs by 70%

- Company: SiliconANGLE (siliconangle.com)
- Announced: 2026-10-19T22:25:50+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://siliconangle.com/2026/10/19/minimax-launches-m3-1-flash-model-says-it-cuts-ai-agent-costs-by-70/
- Record: https://forck.live/items/24800-minimax-launches-m3-1-flash-model-says-it-cuts-ai-agent-costs-by-70
- Subject: Reporting
- Models affected: M3.1 Flash, M3, V4.1-Flash, Space Bunny
- Context window: 1 million tokens
- Pricing: 15 cents per million input tokens; output 60 cents per million tokens; cached input 1.5 cents

MiniMax released M3.1 Flash, a faster and cheaper update to the M3 large language model it launched in June, aimed at the cost of running AI agents that need dozens of model calls per job. MiniMax says agentic tasks cost 70% less to run than on its previous model and inference is close to three times as fast, with output at everyday peak topping 200 tokens per second. The developer guide for the preview release lists 428 billion parameters, barely changed from M3, and credits the extra speed to more aggressive sparsity at inference time. Reasoning stays on for every request, with developers choosing one of five effort levels from Low to Max, and the model accepts text, image and video input with prompts running to 1 million tokens. In MiniMax's own testing, M3.1 Flash scored 61.7 on OSWorld V2, 76.2 on BankerToolBench and 82% on CyberGym, a benchmark built by University of California, Berkeley researchers to test whether AI agents can reproduce real software vulnerabilities. At a flat 15 cents per million input tokens, it lines up against DeepSeek's V4.1-Flash, which charges the same rate off-peak; M3.1 Flash managed 73.7% on DeepSWE v1.1 against DeepSeek's 77%, and neither result has been checked independently. M3.1 Flash is available through the MiniMax API, the M Plan subscription and the MiniMax Code desktop app, with output at 60 cents per million tokens and cached input at 1.5 cents; MiniMax said it will release the Flash weights later and a more capable Pro model is also coming. The launch also confirms that Space Bunny, an anonymous model that appeared on OpenRouter in late September, was an early M3.1 Flash build, with more than 200 trillion tokens going through it while its maker stayed unnamed.

## Evidence

Verbatim from https://siliconangle.com/2026/10/19/minimax-launches-m3-1-flash-model-says-it-cuts-ai-agent-costs-by-70/:

> Agentic tasks cost 70% less to run than on its previous model, the company said, and inference is close to three times as fast, with output at everyday peak topping 200 tokens per second.

---

Record: https://forck.live/items/24800-minimax-launches-m3-1-flash-model-says-it-cuts-ai-agent-costs-by-70
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md