Was checking the opencode-go provider data against the docs at https://opencode.ai/docs/go and found a few things off.
1. Provider doc link goes to the wrong page
providers/opencode-go/provider.toml has doc = "https://opencode.ai/docs/zen". The Go subscription has its own docs page now at https://opencode.ai/docs/go.
2. MiniMax M2.7 is missing cache_write pricing
The docs list cache_write at $0.375/M for M2.7 but the TOML only has cache_read = 0.06 and no cache_write field at all.
3. MiniMax M2.5 has wrong cache_read and missing cache_write
cache_read is set to 0.03 but the docs say $0.06/M. And cache_write = 0.375 is missing from the TOML. Looks like cache_read might have been halved by mistake.
The rest checks out. All 14 active model IDs match. The dual-package setup (provider default @ai-sdk/openai-compatible, per-model override to @ai-sdk/anthropic for MiniMax and Qwen models) lines up with the endpoint table. Pricing on everything else matches. The three deprecated models are marked correctly.
Was checking the opencode-go provider data against the docs at https://opencode.ai/docs/go and found a few things off.
1. Provider doc link goes to the wrong page
providers/opencode-go/provider.tomlhasdoc = "https://opencode.ai/docs/zen". The Go subscription has its own docs page now athttps://opencode.ai/docs/go.2. MiniMax M2.7 is missing cache_write pricing
The docs list cache_write at $0.375/M for M2.7 but the TOML only has
cache_read = 0.06and nocache_writefield at all.3. MiniMax M2.5 has wrong cache_read and missing cache_write
cache_readis set to0.03but the docs say $0.06/M. Andcache_write = 0.375is missing from the TOML. Looks like cache_read might have been halved by mistake.The rest checks out. All 14 active model IDs match. The dual-package setup (provider default
@ai-sdk/openai-compatible, per-model override to@ai-sdk/anthropicfor MiniMax and Qwen models) lines up with the endpoint table. Pricing on everything else matches. The three deprecated models are marked correctly.