Qwen4 Max Launches on Together AI at 30% Discount
The non-obvious point is that a 2.4T MoE with only 71B active parameters can be served profitably by third-party providers, signalling aggressive price competition in the agentic-model market. Developers building coding agents or long-context pipelines should benchmark it against Claude and GPT alternatives while the 30% discount holds. Teams in ASEAN and China markets gain another frontier-tier option outside US-only ecosystems.