novita-ai: add 8 new models and remove 2 deprecated models

[Alex-wuhu]

Summary

Sync Novita AI model listings with the live API (https://api.novita.ai/openai/v1/models).

Added 8 models:

• zai-org/glm-5.1 — GLM-5.1 reasoning model ($1.40/$4.40 per Mt)
• google/gemma-4-26b-a4b-it — Gemma 4 26B MoE, multimodal ($0.13/$0.40 per Mt)
• google/gemma-4-31b-it — Gemma 4 31B dense, multimodal ($0.14/$0.40 per Mt)
• minimax/minimax-m2.7 — MiniMax M2.7 reasoning model ($0.30/$1.20 per Mt)
• minimax/minimax-m2.5-highspeed — MiniMax M2.5 speed-optimized variant ($0.60/$2.40 per Mt)
• qwen/qwen3.5-27b — Qwen3.5-27B multimodal ($0.30/$2.40 per Mt)
• qwen/qwen3.5-122b-a10b — Qwen3.5-122B MoE multimodal ($0.40/$3.20 per Mt)
• qwen/qwen3.5-35b-a3b — Qwen3.5-35B MoE multimodal ($0.25/$2.00 per Mt)

Removed 2 deprecated models (no longer served by the API):

• kwaipilot/kat-coder
• skywork/r1v4-lite

Details

• All pricing data sourced from the Novita AI LLM pricing page
• Model capabilities (reasoning, tool_call, modalities, etc.) cross-referenced with existing entries on other providers
• After this change, the repo lists 94 Novita AI models, matching the live API exactly

Test plan

• bun validate passes with no errors

🤖 Generated with Claude Code