Skip to content

0.55.7 — blockrun_modal reserves the gateway's new sandbox prices - #211

Merged
VickyXAI merged 1 commit into
mainfrom
fix/modal-pricing-by-size
Nov 6, 2026
Merged

VickyXAI merged 1 commit into
mainfrom
fix/modal-pricing-by-size

Conversation

@VickyXAI

@VickyXAI VickyXAI commented Nov 6, 2026 •

Copy link
Copy Markdown
Contributor

A create at 300 seconds or less, with 0.125 to 1 CPU and at most 1024 MB, pays the flat rate. Any other create pays the hourly rate times the requested lifetime, upfront, never below the flat rate. Hourly: CPU $0.25, T4 $4, L4 $4.50, A10G $5, A100 $7, H100 $9. Flat rates do not change. cpu below 0.125 is refused, in the gateway and the wrapper.

The estimator ignored cpu and memory and held the old hourly rates. With the new gateway prices, {gpu:"T4",timeout:86400} would reserve about $36 and sign $96.001, and {gpu:"T4",cpu:8,memory:32768,timeout:300} would reserve $0.052 and sign $0.334. The budget cap and confirm dialog both read this estimate.

The skill's GPU example also dropped a custom image and setup_commands, which the gateway refuses.

Verification. Full suite and tsc green. Removing the size check fails the cost test.

Release order. Ship after the wrapper deploy and before the gateways: BlockRunAI/blockrun#1000, BlockRunAI/blockrun-sol#572, BlockRunAI/blockrun-arc#67, BlockRunAI/blockrun-nano#40. Until they deploy, this version over-reserves on hourly creates, which is the safe direction.

🤖 Generated with Claude Code

@VickyXAI
VickyXAI merged commit 6e3b486 into main Nov 6, 2026
3 checks passed
@VickyXAI
VickyXAI deleted the fix/modal-pricing-by-size branch November 6, 2026 21:58
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant