Artem Senenko

DeepSeek v4 Flash has received an incredible upgrade and is now very close to Opus 4.8.

The new version is already available via API and has received Responses API support, as well as full Codex compatibility.

The model has improved significantly in benchmarks:

> DeepSWE — 54.4%, surpassing GLM-5.2 and its own version, 4 Pro, by a wide margin, almost reaching the level of Opus 4.8.
> TerminalBench — 82.7%, surpassing 4 Pro and GLM-5.2, and again delivering results almost on par with Opus 4.8.

OpenAI Challenges Cheap Intelligence from China:

OpenAI has reduced the price of GPT-5.6 Luna by 80%, to $0.20 per million input tokens and $1.20 per million output tokens.

Terra has dropped in price by 20%, to $2/$12, respectively. Luna is now 25 times cheaper than Sol.

The changes affect both API pricing and paid subscription usage.

And for GPT-5.6 Sol, a Fast mode has been added to the API. This provides up to 2.5x acceleration at double the price—a 25% better speed-to-cost ratio for the same level of intelligence.

5 month ago builded console tool that can monitor a Claude Code session, and I'm proud that is still useful

here is how the vote engine in my uptime monitor works.

one bad location can never page you. that is the whole point.

uptimepage.dev/blog/stop-false-uptime-alerts

Claude Opus 5 is out now.

It outperforms Fable 5 in almost every test, yet costs half as much.

Here’s an interactive map of how Uptimepage works:

choose one runtime workflow and it lights up live across every layer.

No slidedeck, no photos from whiteboards.

Kimi-3 better in WebDev task than Claude Fable

Do you still use Claude Fable?

i made a small free tool for people who work with SLOs.

it is an error budget + burn rate calculator. you put in your target uptime and what you really got, and it shows how much downtime budget you have, how much is left, how fast you are using it, and the point it runs out. no signup, nothing to install. try it.

uptimepage.dev/tools/error-budget-calculator

My disk hit 100% full in production. Postgres crashed. But my monitor lost zero customer check results, retries caught every failed write.

Then I found the cause. ClickHouse was quietly writing gigabytes of logs about itself, mostly system.text_log and trace_log. Logs I never asked for.

The fix: truncate the log tables (12 GB back), then a config file that caps them for good. I also added a disk alert at 80%, so next time I get days of warning, not minutes.

Full postmortem and the copy-paste config:

uptimepage.dev/blog/clickhouse-system-tables-fi…

wow, this guy really know how to work with Claude Fable 5

source available: github.com/AbhishekBadar/portfolio

have you ever heard about email bombing attack?
uptimepage.dev/blog/email-bombing-uptime-pages

one small task and Fable 5 used almost all tokens in one hour 😬

Claude Fable 5 is back! we still see some problems with connection, perhaps because the influx of people who want to try app.uptimepage.dev/m/xLfNohjJIBzfUCu6bMEcXCdgty…

Claude Fable 5 is still unavailable. you can check it on uptimepage.dev

app.uptimepage.dev/m/xLfNohjJIBzfUCu6bMEcXCdgty…

still waiting while new Fable 5 will be enabled? subscribe for updates to be first when it will be ready
claude-status.uptimepage.dev

hey there, I just shipped uptime monitoring tools, it's open source github.com/uptimepage/uptimepage

if you want to have good observability of your service and be notified when service goes down, please join.
it's almost free for all basic use-cases.

also want to heard your feedback, what is good and what is bad. if you need some specific features then ping me in DM