Models
Every frontier model. One teammate.
Henry is model-agnostic. Pick a preset and never think about it again, or choose the exact model and reasoning level, per workspace, changeable anytime.
Presets
Three good defaults, zero tuning
Ultra
Understands what you actually mean with the least hand-holding, and stays sharp on the longest, most complex work. The closest thing to a real teammate.
Smart
Handles most day-to-day work well. Very complex or ambiguous asks land better on Ultra.
Balanced
Anthropic's fast everyday model. Strong on writing and careful instruction following, a good middle step when you want a Claude model without Ultra's price.
Fast
The quickest and cheapest option. Great for short questions, lookups, and simple tasks; hand the complicated work to Smart or Ultra.
Full catalog
Use whatever model you want
The latest from Anthropic and OpenAI, side by side. Switch per workspace, anytime. Your choice is always explicit, never silently rerouted.
Claude Fable 5.1
Anthropic's newest and most capable model, the successor to Fable 5 for the hardest reasoning and longest autonomous work.
GPT-6 Astra
OpenAI's frontier GPT-6 model for the most demanding reasoning, agentic, and long-context work.
Claude Fable 5
Previous-generation Fable, still excellent for demanding reasoning and long-running autonomous work.
GPT-5.6 Sol
OpenAI's flagship GPT-5.6 model.
Claude Opus 5.5
Newest and strongest Opus model for coding, agents, and complex workflows, priced below Opus 5.
Claude Opus 5
Previous-generation Opus, still strong for coding, agents, and complex workflows.
GPT-5.6 Terra
GPT-5.6 tier with strong performance at a lower price than Sol.
Claude Sonnet 5.5
Newest Sonnet, the best balance of speed and intelligence for everyday work, at Sonnet 5's price.
GPT-5.6 Luna
Efficient GPT-5.6 tier for high-volume work.
Grok 4.7
xAI's frontier model for agentic coding and professional knowledge work, built for complex multi-hour tasks.
Claude Sonnet 5
Previous-generation Sonnet: near-Opus quality on most tasks, faster and cheaper.
Claude Opus 4.8
Older-generation Opus model for coding, agents, and complex workflows.
Grok 4.6
Previous-generation Grok, built for long-running agents, agentic coding, and interactive work.
Grok 4.5
Previous-generation Grok model, strong on coding, knowledge work, and STEM.
Gemini 3.1 Pro Preview
Google's most capable Gemini model for complex reasoning, coding, and multimodal work. Preview release.
Gemini 3.8 Flash
Google's fast, low-cost Gemini model for high-volume work, with a 1M-token context window.
Kimi K3
Moonshot AI's open flagship for agentic work and coding, with a huge context window.
Kimi K3 Fast
Kimi K3 on a faster serving path — the same model with lower latency at a higher per-token price.
Claude Sonnet 4.6
Fast, capable older-generation Sonnet.
Claude Haiku 5.5
Anthropic's fastest and cheapest model, a big step up from Haiku 4.5 on tool use and agent work. Best for quick, high-volume tasks.
Claude Haiku 4.5
Previous-generation Haiku. Runs without extended thinking.
GPT-5.5
Previous-generation OpenAI model, superseded by the GPT-5.6 family.
$ to $$$$ compares typical usage cost with the Smart preset ($). Hover the dollar signs for the exact estimate. You're billed on actual usage.
Our philosophy
Your model, your call
Switch anytime
Change models per workspace whenever you like. New frontier models land in the catalog as they ship.
No silent failover
Henry uses the model you picked, full stop. If it's unavailable, you hear about it. You're never quietly served a cheaper model.
Bring your own key
On Enterprise, bill Anthropic or OpenAI usage through your own account instead of Henry's.