Maxwel Agents
Managed Agents
Anthropic Claude Opus 5.5
Multi-step agent workflows and deep multi-file analysis
Anthropic Claude Sonnet 5.5
Daily driver for code reviews and general tasks
Anthropic Claude Fable 5.1
Mythos-class flagship for the longest autonomous tasks and deepest analysis
Anthropic Claude Haiku 5.5
Very cheap and very fast — for tool use, subagents and high-throughput pipelines
OpenAI GPT 6.1
Frontier model for agentic workflows with native tool orchestration and multimodal inputs
Google Gemini 4 Carbon
Google's most capable model, via Vertex AI in the EU multi-region
Google Gemini 3.8 Flash
Google's fast workhorse, via Vertex AI in the EU multi-region
DeepSeek DeepSeek V4.1 Flash
DeepSeek's new standard model with image understanding — ahead of V4 Pro on most benchmarks per the vendor, at Flash speed with 1M context
DeepSeek DeepSeek V4 Flash
Compact MoE variant for high-volume throughput with very long context at minimal cost
Alibaba Qwen 3.8 27B
Open everyday model with image understanding and 262K context — hosted in the EU
Alibaba Qwen 3.8 Flash Next
The fastest model in the lineup: multimodal MoE with just 6B active parameters, built for agents, coding and long-running office workflows
Alibaba Qwen 3.8 Max
Open frontier model with 2.4 trillion parameters for coding, research and long-horizon agents — the only model of its class available with EU data protection
Moonshot Kimi K3
Agentic orchestration with long context for multi-step workflows
Z.ai GLM 5.3 Flash
Natively multimodal open-source model (MIT) with 1M context at the lowest price — understands images too
Z.ai GLM 5.3
Top-tier model for agentic engineering and SWE-Bench tasks
Google Nano Banana 2.1
Fast image generation and editing for iterative workflows — up to 14 reference images, half the price of Nano Banana 2
Google Nano Banana Pro 2
Highest detail accuracy for complex scenes with many subjects
Google Vertex AI Veo 3.1 Fast
Google's video model via Vertex AI on the gateway — cinematic motion with sound, up to 4K
ElevenLabs ElevenLabs TTS v3
Studio quality for audiobooks and voiceovers in 70+ languages
OpenAI Whisper Large v3
Speech-to-text in 99 languages with high robustness
ElevenLabs ElevenLabs Scribe
Transcription with the best word accuracy — speaker diarization, 99 languages, per-word timestamps
Brave Software Brave Search
Seat and processing in the US (AWS)
Perplexity AI Perplexity Search
Seat and processing in the US, own index
Mojeek Ltd Mojeek Search
Seat and processing in the United Kingdom (adequacy decision), own index