feat(wafer.ai): add DeepSeek V4 Pro

[Sewer56]

Add DeepSeek V4 Pro to Wafer

Summary

Adds DeepSeek-V4-Pro model definition to the wafer.ai provider.

Details

• File: providers/wafer.ai/models/DeepSeek-V4-Pro.toml
• Family: deepseek-thinking
• Context: 1,000,000 tokens
• Max output: 384,000 tokens
• Reasoning: yes (thinking mode with interleaved reasoning_content)

[Sewer56]

CC @rekram1-node

I'm not seeing the max reasoning option here on v1.14.29 of OpenCode.

// transform.ts:576
if (model.api.id.includes("deepseek-v4")) {
  efforts.push("max")
}

It seems the check inside OpenCode is case sensitive. Would you be able to get a quick fix through in there?

[rekram1-node]

@Sewer56 yeah I can do a .toLower

[rekram1-node]

Can we use the "extends" format to reduce duplicating so much of the definition (it's a newer concept we introduced)

[Sewer56]

Yeah sure, will get it sorted when back at desk.

[Sewer56]

@rekram1-node All sorted 👌

[Sewer56]

Thanks, that'd work 👌

[rekram1-node]

anomalyco/opencode#25012