Viewpoint
DeepSeek Flash leads by token volume due to cheap pricing
DeepSeek Flash leads by token volume due to cheap pricing
OpenCode's published data showed DeepSeek Flash leading by token volume, with GLM close on unique users. Because DeepSeek Flash is very cheap, many users switch to it to extend work after hitting limits, shaping usage patterns differently from typical Western assumptions.
- Speaker
- Jay V
- Source timestamp
- 11:02
More from this interview
- OpenCode grew to 13 million monthly users
- Anthropic's block boosted OpenCode's visibility
- OpenCode aims to make coding agents affordable globally
- Users moved from Claude Code to open source models
- KIMI models overtook frontier models in OpenCode usage, prompting a paid subscription.
- China is OpenCode's largest usage region at 17 percent
- Large companies use OpenCode to avoid lock-in and preserve flexibility.
- Enterprise demand grows from existing users and asks
- Ramp's Slack bot points to new coding agent uses
- OpenCode's economics rely on volume discounts and subsidies
- OpenCode acts as a model marketplace where vendor lock-in harms users
- Growth tied to improvements in the open source model market
- Model labs will specialize in different niches
- Global usage yields stable GPU utilization, improving unit economics
- OpenCode's name reflects its open alternative position and models.dev support
- Modern terminal experience built for core developers and open source community
- Same legal entity persisted through many attempts before OpenCode success