Logicplanes

[Kaeso @kaesoapp]

Fast models deserve a fast gateway. 0.5.6 also translates streamed replies 1.3 to 1.6x faster and decodes large requests about a third faster.

https://t.co/Aj1AoN1xtd

[Kaeso @kaesoapp]

Cerebras is now a built-in provider in Kaeso.

Add your key and call Cerebras models from any app format: Chat Completions, Responses or Messages.

Out now in v0.5.6.

[Devin Oldenburg @dvnold]

0.5.0 is the first Kaeso release where the model data is ours: signed KPMR catalogues instead of https://t.co/sQ5nawNGf5. Plus a response cache and optional ClickHouse.

[Logicplanes @logicplanes]

Kaeso 0.5.0 is out: a response cache, optional ClickHouse, and signed model data from our own registry.

[Kaeso @kaesoapp]

Shipped Kaeso v0.5.0

Ask twice, pay once: projects can now turn on a response cache. A repeated identical request gets the stored reply, at no provider cost.

[Kaeso @kaesoapp]

Shipped Kaeso v0.4.8

Four new built-in providers: Moonshot (Kimi), QwenCloud, Cohere and Ollama Cloud.

Add a key and call their models from any app format.

[Kaeso @kaesoapp]

Also in 0.4.8: smoother dialogs, and custom providers and SSO discovery can never reach Alibaba Cloud's metadata service.

Full changelog: https://t.co/hLOL1S1Sk5

[Kaeso @kaesoapp]

Your app speaks Chat Completions, Responses or Messages. Kaeso translates, so Kimi K3, Qwen3 Max, Command A Plus or GLM-5.3 work with the SDK you already use.

QwenCloud brings its embeddings too.

[Logicplanes @logicplanes]

New in Kaeso: Kimi, Qwen, Cohere and Ollama Cloud, built in.

[Kaeso @kaesoapp]

Shipped Kaeso v0.4.8

Four new built-in providers: Moonshot (Kimi), QwenCloud, Cohere and Ollama Cloud.

Add a key and call their models from any app format.

[Kaeso @kaesoapp]

Plus console polish: status never shown by colour alone, Undo when you disable a key, clear messages when something can't load, and better phone layouts.

Fresh installs start with the model catalogue of 30 October 2026.

Full changelog: https://t.co/W5WfznZET1

[Kaeso @kaesoapp]

Codex sessions with subagents now work on Grok and every other non-OpenAI model. A tool_choice sent without tools also no longer reaches providers that refuse it.

[Kaeso @kaesoapp]

Gateway fixes: keys now expire on the date you chose, a replaced provider key stops being used right away instead of up to 30 seconds later, and invitations send one address at most five emails an hour.

[Kaeso @kaesoapp]

Shipped Kaeso v0.4.7
@vercel AI Gateway is now a built-in provider. Add its key and call its models from any app format Kaeso speaks: Chat Completions, Responses or Messages.

[Kaeso @kaesoapp]

Shipped Kaeso v0.4.6

Kaeso overhead is now measured more accurately: it no longer counts the time your app takes to upload the request body. The console also stops logging a React error while pages load, and fresh installs start with the model catalogue of 28 October 2026.

Full changelog: https://t.co/d2gvFOjL9C

[Kaeso @kaesoapp]

Claude's calls to tools without parameters now reach Codex and Chat apps as {}, and tool calls whose arguments aren't an object no longer stop a thread.

Fresh installs start with the model catalogue of 27 October 2026.

Full changelog: https://t.co/TPtdZZR5wm

[Kaeso @kaesoapp]

Every control in the console and the Kaeso UI library now shows its hover, press and focus states, keeps focus where it belongs, works with the keyboard, mirrors right to left and stays readable in forced colours.

[Kaeso @kaesoapp]

the full transition will probably take happen as we release v0.5.0

[Kaeso @kaesoapp]

The request log now filters by format and shows each request's cached share. Usage reports a cache hit rate per route, so you can see which apps actually benefit from prompt caching.

[Kaeso @kaesoapp]

Shipped Kaeso v0.4.5

You can now see how much of your traffic is cached, the whole console got an accessibility pass, and Claude works better behind Codex. Details below.

[Devin Oldenburg @dvnold]

Between you and the world.