AI models on VM0

Claude Opus 4.8

Anthropic's newest Claude 4 flagship. Leads SWE-bench Pro and OSWorld-Verified at release, with dynamic workflows and a 3× cheaper fast mode.

Claude Opus 4.7

Anthropic's flagship Claude 4 model. Highest reasoning quality in the family on multi-step agent loops, long-context recall, and code edits.

Claude Opus 4.6

Anthropic's previous flagship. Same credit cost as Opus 4.7. Keep it pinned only if a specific agent has been validated against this version.

GPT-5.5

OpenAI's flagship reasoning model. Top-tier coding and tool-use, routed through the Codex framework on VM0.

Claude Sonnet 4.6

The default for most VM0 agents. Strong tool-routing accuracy, good long-context behaviour, and the credit baseline. Every other model is priced relative to Sonnet 4.6.

GPT-5.4

OpenAI's workhorse model. ×1 credits, multimodal vision, strong code and tool use — the default GPT-5 for most agents.

GLM-5.1

Z.AI's flagship. Up to a 1M-token context window. Strong for whole-codebase or whole-knowledge-base agents at well below Sonnet pricing.

Claude Haiku 4.5

The fast, cheap Claude. Good enough for routing, short summarisation, and simple tool calls at a fraction of Sonnet's cost.

GPT-5.4 Mini

OpenAI's cost-saving GPT-5. ×0.3 credits, fast and multimodal — the right pick for bulk pre-filter work on the Codex framework.

Kimi K2.6

Moonshot's latest. Best-in-class long-context recall in our internal evaluation and a Claude-compatible interface.

DeepSeek V4 Pro

DeepSeek's flagship. Strong reasoning at one-third of Sonnet's credit cost, Claude-compatible API.

Kimi K2.5

The previous Kimi generation. Cheaper than K2.6 but with weaker tool-use; pin it only if a specific agent was validated on this version.

MiniMax M2.7

Strong multilingual reasoning at one-tenth of Sonnet's credit cost. Generous timeout for long thinking steps.

DeepSeek V4 Flash

The cheapest model in the lineup. 50× less than Sonnet 4.6. Good for high-volume single-shot tasks where the prompt does most of the work.

GPT Image 2

OpenAI's newer image model for generation and editing, with flexible sizing and stronger prompt adherence than GPT Image 1.

GPT Image 1

OpenAI's text-to-image model. Strong on stylised illustration, character work and editing.

Black Forest Labs' top-tier image model. Highest aesthetic ceiling in the curated lineup — use it for hero shots and editorial-grade output.

ByteDance's SeedDream 4. Photoreal aesthetic at a low per-image cost — the natural default when the brief is a photo.

Veo 3.1 Fast

Google's text-to-video model with native audio. Fast generation, cinematic style and synchronised soundtrack in a single pass.

Kuaishou's Kling V3 4K. Stylised aesthetic, long clips up to 15 seconds, 4K output — the ceiling for editorial and anime-style video.

ByteDance's Dreamina Seedance 2.0. Cheapest per-second video in the curated lineup — the right pick for bulk and cost-driven generation.