LLM models

Supported AI models

TrustChat talks to every model through a swappable provider adapter – selectable per bot, with no provider-specific code in the domain model.

Model Provider Status Context window Input $/1M tokens Output $/1M tokens Quality & price positioning
gpt-5.4-mini OpenAI Available 400000 $0.75 $4.50 Standard – excellent price-performance ratio
gpt-5.4-nano OpenAI Available 400000 $0.20 $1.25 Cheapest – for simple FAQ bots
gpt-5.4 OpenAI Available 1050000 $2.50 $15.00 Higher quality – about 3× the cost of mini
gpt-5.5 OpenAI Available 1050000 $5.00 $30.00 Flagship – highest quality, significantly pricier
gpt-5.6-luna OpenAI Available 1050000 $0.20 $1.20 New 5.6 generation – fastest, most cost-efficient tier (Luna)
gpt-5.6-terra OpenAI Available 1050000 $2.00 $12.00 5.6 generation – more quality than Luna, cheaper than Sol
gpt-5.6-sol OpenAI Available 1050000 $5.00 $30.00 5.6 generation – high quality, pricier than Terra
gpt-6-astra OpenAI Coming soon 1000000 $10.00 $50.00 New flagship (GPT-6) – highest quality, significantly pricier, not yet enabled
claude-haiku-4-5 Anthropic Available 200000 $1.00 $5.00 Anthropic – fast & cheap, quality on par with gpt-5.4-mini
claude-sonnet-5 Anthropic Available 1000000 $2.00 $10.00 Anthropic – highest quality for coding & context
claude-opus-5 Anthropic Available 1000000 $5.00 $25.00 Anthropic – more quality than Sonnet 5, pricier
claude-fable-5-1 Anthropic Coming soon 1000000 $10.00 $50.00 Anthropic – most capable model, very expensive, not yet enabled

Prices are the providers' official list prices (USD, as of September 2026) and are meant to guide model choice. The actual cost report in the super-admin area currently bills every tenant off a single, globally configured rate (trustchat.pricing.*) regardless of which model is selected – per-model billing isn't implemented yet. Models marked "Coming soon" were announced recently or still need adapter changes and can't be selected as a bot's model yet.