MiniMax M2.7 uses self-critique training to cut agent costs
10/24/2026 — 10/24, 00:19·1 sources (all)·1 reports (all)
Key takeaways
- MiniMax released M2.7, a model trained with self-critique that completed over 100 rounds of autonomous optimization and now handles 30-50% o…
MiniMax released M2.7, a model trained with self-critique that completed over 100 rounds of autonomous optimization and now handles 30-50% of the reinforcement learning workflow. It scores 56.22% on SWE-Pro and costs $0.30 per million input tokens, far below Claude Opus 4.6 at $15.
Latest turnMiniMax's M2.7 used self-critique across 100+ optimization rounds, handling 30–50% of the RL workflow. It scores 56.22% on SWE-Pro at $0.30 per million input tokens, far below Claude Opus 4.6.
- Heat index
- 80
- All sources
- 1
- All reports
- 1
- First seen
- 8 hours ago
Reports on this story headlines open the original
MiniMax's M2.7 used self-critique across 100+ optimization rounds, handling 30–50% of the RL workflow. It scores 56.22% on SWE-Pro at $0.30 per million input tokens, far below Claude Opus 4.6.
DEV Community · AIAI score 75
Other stories people are talking about
- Heat index 238SurgeMiniMax M3.1 Flash comes to Tenstorrent Cloud with open weights3 sources
- Heat index 809SurgeKimi K3.1 Pro debuts as top open-weight model in agent cyber defense9 sources
- Heat index 715NewSurgeOpenAI resumes frontier training on isolated offline network8 sources
- Heat index 365US banks block AI agents that use customer credentials to access accounts6 sources
- Heat index 345DeepSeek closes around CNY 10B funding round led by CATL and Tencent5 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion