OpenCode Go
$10/mo (Go Plus $40/mo)
Models: 31 models, mostly open-weight (DeepSeek V4.1 Flash, Qwen3.8, GLM-5.3, Kimi K3, MiniMax M3) plus Grok, GPT-6 Luna and Claude Haiku 5.5
What "flat" means: Fixed price with a monthly dollar limit that varies by model (on Go: $15 of GLM-5.3 or Kimi K3, $60 of MiniMax M3, $80 of DeepSeek V4.1 Flash), metered at model token rates.
The catch: Tight windows on every model: 20% of the monthly limit per 5 hours and 50% per week (a $60 model caps at $12 per 5 hours and $30 per week). Premium models such as Kimi K3, Claude Haiku 5.5 and GPT-6 Luna get only $15 a month on Go. Single-member workspaces only, and traffic must look like a coding agent.
Pick OpenCode Go when
+You run mostly open-weight models and want the cheapest entry price ($10/mo)
+Your usage fits inside $30/week (moderate, not always-on)
+You want a plan built by OpenCode that also works with Hermes and OpenClaw
MiniMax M Plan
$22 / $55 / $132 per mo (Go / Explore / Build; annual = 10 months' price)
Models: MiniMax M3.1 Flash Preview for text, plus image and audio models; H3 video on Explore and Build only
What "flat" means: Fixed price with 1x / 3x / 7.5x the Go tier's usage in a 5-hour window and a 7-day window (video weekly only), all modalities sharing one pool, no carry-over. Absolute amounts aren't published.
The catch: The M Plan replaced the Token Plan, which is closed to new purchases (existing subscribers keep it while auto-renew stays on). Window sizes and per-tier concurrency are unpublished, peak hours bring dynamic rate limits, and MiniMax recommends pay-as-you-go for production.
Pick MiniMax M Plan when
+M3.1 Flash Preview plus image, audio and video generation on one subscription fits your work
+Individual interactive workflows that fit the rolling windows
+You want a low flat entry from a single vendor and can pay annually for two months free