Add Nebul provider with 10 chat models

[maximoei]

Summary

Adds Nebul as a new inference provider with 10 chat completion models.

Provider

• Provider ID: nebul
• API: OpenAI-compatible at https://api.inference.nebul.io/v1
• SDK: @ai-sdk/openai-compatible
• Auth: NEBUL_API_KEY
• Docs: https://docs.nebul.io/docs/inference-api/models/model-catalog

Models Added

Model ID	Type	Capabilities
mistralai/Mistral-Large-3-675B-Instruct-2512-NVFP4	extends mistral/mistral-large-2512	Chat, Image, Tools, JSON
Qwen/Qwen3-30B-A3B-Instruct-2507	extends alibaba/qwen3-coder-30b-a3b-instruct	Chat, Text
openai/gpt-oss-120b	extends amazon-bedrock/openai.gpt-oss-120b-1:0	Chat, Tools, Reasoning
Qwen/Qwen3-VL-235B-A22B-Thinking	extends alibaba/qwen3-vl-235b-a22b	Chat, Image, Tools, Reasoning
meta-llama/Llama-4-Maverick-17B-128E-Instruct	extends llama/llama-4-maverick-17b-128e-instruct-fp8	Chat, Image, Tools
Qwen/Qwen3-VL-235B-A22B-Instruct-FP8	extends alibaba/qwen3-vl-235b-a22b	Chat, Image, Tools, JSON
mistralai/Devstral-2-123B-Instruct-2512	extends mistral/devstral-2512	Chat, Image, Tools, JSON
Qwen/Qwen3.5-397B-A17B	extends alibaba/qwen3.5-397b-a17b	Chat, Image, Video, Audio, Tools, Reasoning, JSON
zai-org/GLM-5.1-FP8	extends zai/glm-5.1	Chat, Tools, Reasoning, JSON
MiniMaxAI/MiniMax-M2.5	extends minimax/MiniMax-M2.5	Chat, Reasoning, Tools

Notes

• All models use extends to reference canonical sources, with overrides for Nebul-specific capabilities
• openai/gpt-oss-120b overrides reasoning = true with reasoning_options and interleaved field, since Nebul's deployment supports reasoning (unlike the Bedrock canonical)
• [cost] sections are omitted (Nebul does not publish public pricing)
• Logo uses currentColor for theme compatibility
• Context limits match Nebul's documented specs (may differ from canonical sources)