Skip to content
AGNT
All signals
Proof·4 min

Real token costs: Haiku $0.80/1M vs Sonnet $3.00/1M

Claude Haiku 4.5 powers free and starter tier conversations at $0.80 input / $4.00 output per million tokens. Claude Sonnet 4.6 powers pro tier at $3.00 / $15.00 per million tokens. Max output: 1,024 tokens on Haiku, 2,048 on Sonnet.

Token costs in production. These are the real numbers from llm_gateway.py, not marketing approximations.

Claude Haiku 4.5: $0.80 per million input tokens, $4.00 per million output tokens. Max output capped at 1,024 tokens per response. Used for free tier, starter tier, and venue starter tier conversations. Handles the bulk of volume.

Claude Sonnet 4.6: $3.00 per million input tokens, $15.00 per million output tokens. Max output capped at 2,048 tokens per response. Used for pro tier, venue growth, and venue pro tier conversations. Reserved for complex queries.

At ~500 tokens per message on average, Haiku costs approximately $0.0024 per message round-trip. Sonnet costs approximately $0.009 per message round-trip. Pro tier users get unlimited Sonnet at $29/month — covering approximately 3,200 Sonnet messages per month before the margin inverts.

Global daily budget cap: $500 across the entire platform. If all users collectively burn $500 in a single day, new LLM requests return a graceful 'budget exceeded' response. This has never triggered in production but exists as a safety net.

Share this signal

Submit to

Public submit links. No API keys. Opens in a new tab with the title and URL pre-filled.

Copy and paste

Build on the same network.

Every signal comes from a system you can build on today.