MiniMax more than doubles sales - here's what it means for sales teams
MiniMax posted a 159% jump in 2025 revenue to US$79 million, beating the US$71.4 million estimate. Demand is clear: buyers want low-cost, open-source models that can compete with premium US options.
The flip side: net loss widened to US$1.87 billion, driven largely by fair value losses on financial instruments and the heavy costs of model training and user acquisition. This was the company's first earnings report since raising US$600 million in its Hong Kong IPO.
The numbers that matter
- Revenue: US$79 million for 2025 (+159% year over year), above the US$71.4 million consensus.
- Losses: Net loss of US$1.87 billion vs. US$465.2 million prior year, with the bulk tied to fair value losses on financial instruments.
- Market position: Shares have more than quadrupled since listing; valuation sits around US$30 billion, approaching names like JD.com and Kuaishou in market cap.
- Product: The M2.5 model is smaller and cheaper; while it trails top open-source peers from Alibaba and Zhipu per Artificial Analysis, it's the most-used model on OpenRouter on a weekly basis.
- Competition: Independent builder in China competing with Zhipu, ByteDance, and Alibaba; new releases are landing ahead of a much-watched DeepSeek upgrade.
- Scale gap: MiniMax remains small against US leaders. OpenAI's 2025 annualised sales reportedly topped US$20 billion, with a much higher valuation.
Why sales leaders should care
Cheaper, "good-enough" models are changing deal math. For many workflows - prospecting, list research, call prep, summarising discovery notes - cost per output now beats chasing the absolute best benchmark score.
Distribution matters too. If M2.5 is ranking high on OpenRouter, your team can test and swap models without heavy engineering. That creates room to move budgets from licenses into volume and experimentation.
Practical moves for your team
- Pilot low-cost models where quality tolerance is higher: outbound drafts, enrichment, meeting summaries, objection libraries. Route premium tasks (proposal finals, enterprise emails) to higher-end models.
- Run a model mix: Keep 2-3 interchangeable options (e.g., MiniMax M2.5, Alibaba's open-source, a US model). Avoid lock-in and use price competition to your advantage.
- Calculate total cost: Don't judge on token price alone. Track latency, context limits, re-run rates, tool usage, and prompt maintenance time. Price caps and usage alerts prevent budget creep.
- Set guardrails: With recent allegations around model training and IP, require data-provenance warranties, logging, and indemnities. Get security reviews for PII, call notes, and CRM data.
- Benchmark on your workflows: Public leaderboards help, but your SDR and AE tasks are the truth. Score outputs on accuracy, tone, and conversion impact - not just BLEU or MMLU.
How to run a fast bake-off (one week)
- Pick 4 tasks: cold email first draft, lead enrichment, call summary → next-step draft, proposal outline.
- Test three models: your current default, MiniMax M2.5, and one competitor. Keep prompts, datasets, and scoring criteria identical.
- Score and measure: precision, hallucinations, tone fit, time saved per rep, cost per 100 tasks, and downstream conversion signals (opens/replies/meetings) on a small live slice.
- Decide routing: choose a default + a cheaper fallback for bulk work. Add an "upgrade" route for high-stakes accounts.
What to watch next
- Pricing pressure: If DeepSeek and peers push updates, expect more discounting - good for budgets, tough for vendor stability.
- Enterprise readiness: SLAs, observability, SOC 2, regional hosting, and CRM-native integrations will separate pilots from full rollouts.
- Content tools: MiniMax's consumer side (AI avatars, Hailuo video) could unlock low-cost creative for outbound and ABM - keep an eye on export quality and brand safety.
- Profit path: Rising losses vs. user growth. Favor contracts with clear service credits, uptime commitments, and pricing protections.
Want a structured way to upskill your team on model selection, routing, and AI-driven prospecting? Explore our AI for Sales resources.
Your membership also unlocks:
