polytoken dot dev

Polytoken stable just advanced to 0.6.13. Lots of bug fixes, some enterprise-friendliness adjustments, new model providers, new models, better compaction, better post-compaction, shell mode with !, bunch of other stuff. `curl -fsS get.polytoken.dev | bash` is friend. docs.polytoken.dev/changelog/

so this is how @ed3d.net ingests tickets for @polytoken.dev while he's out and about

New video up! Kind of a follow-on from the last one, because a YouTube comment made me feel bad for the author and I think a lot of people are buying into the mindset that it conveyed. Things wander from there. Cool jackets are mentioned. Next one is a cybersecurity talk. youtu.be/a7gxS7w5gi0

i basically agree. my multi-model setup right now is openai luna and sol, glm 5.2 (soon to be 5.3!!), kimi k3, and deepseek v4. i removed all anthropic models from my menu.

Tool flows are Extremely Experimental and the agent isn't biased towards using them, but I think there are pretty good use cases here, especially when coupled with datastores and other stuff that might dump a lot of text your agent doesn't actually care about.

Polytoken 0.8.0 released to latest, with a new feature: tool flows. Agents can orchestrate the inputs and outputs of their tool library in a sandboxed environment and return to the harness only what it needs. Stable has updated to 0.7.4. docs.polytoken.dev/changelog

Tree shaking must feel so good

polytoken.dev is now a website plase clap

On Max, GLM-5.3's token usage is best described as "bullshit" and the high/max performance divide seems very small. Polytoken defaults both 5.2 and 5.3 to high, because I am ungovernable by model provider advice.

do I dare make a polytoken release that is 0.6.7 do I want that smoke

📣 maaaaany people are saaaaaaaying 📣

someone called a thing deepinfra when derpthra was right there

PSA - Polytoken stable updating to 0.5.10. Whole bunch of fixes and improvements that I think you'll like, you just get to like them later than the people on the latest track docs.polytoken.dev/changelog/

(psst. I’m not posting releases because now you don’t have to do anything to get them with the autoupdater. But there have been five so far since the switch)

Good news: Polytoken 0.5.1+ has an updater! Bad news: you have to go get it. Homebrew's deprecated now. Uninstall first. Once you reinstall with the manual download or the script, you'll be on the update track. curl -fsS get.polytoken.dev | bash ... ...I'm gonna leave the preview there, enjoy

Polytoken 0.5.0 is out on stable! Whole bunch of new features, perf improvements, and fixes, check the changelog for details. brew install polytoken/tap/polytoken{,unstable} docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.10 is out! Kimi Code/Moonshot are now built-in providers. Also adding auto crash reporting (`config ui` to opt out), `/feedback` for easier shipping of logs etc, and improved subagent tooling. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.9 is out! Improvements to compaction quality, supporting ~/.agent/skills, and a fix to autonomous permissions mode with GPT 5.6 models. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.8 is out! Small fixes and a couple UX requests -- you can now look behind ask_user_question and handoff_plan interrogatives to read the conversation history behind them. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.7 is out! GPT-5.6 Luna is good enough to be both a mini and a full model, so now you can just...do...that. Highlight model in the config UI! Press D! Go "ooooooh". brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.7 is out! GPT-5.6 Luna is good enough to be both a mini and a full model, so now you can just...do...that. Highlight model in the config UI! Press D! Go "ooooooh". brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.6 is out! A pile of performance improvements and additional tracking for some odd behaviors from GPT-5.6 on the Responses API. This IS a more experimental build than usual, download emptor. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.4.2 and 0.5.0-unstable.5 are out - fixes a bug in the GPT 5.6 provider implementation. update yr bookmarks accordingly

Polytoken 0.4.1 and 0.5.0-unstable.4 are out! Adds GPT-5.6 support and brings bug fixes/perf to stable, while updating providers and improving /goal on unstable. brew install polytoken/tap/polytoken{,-unstable} docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.2 released! Mostly bug and performance fixes, such as "you can open it in a macOS home directory and not explode in a hail of sandbox errors". We're getting there. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.5.0-unstable.1 released! First steps towards A Cool New Feature that I'm not gonna talk about much...yet. But also a metric ton of perf improvements on long sessions and a couple small bug fixes. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.4.0-unstable.7 out. I know I said .6 would be the last, but I ran into a couple macOS bugs and also had a theme provider to ship, so now you can decorate and wallpaper to your heart's content. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

Polytoken 0.4.0-unstable.6 out. Adds Neuralwatt as a provider, addresses a few weird random issues that I wanted to take care of before 0.4.0. Which, I think, will probably be tomorrow (based on unstable.6). brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation/

here's a promise: polytoken will never decide your input is so unimportant that after sixty seconds it will just jam onward with whatever the hell it wanted to do perhaps the fable huffers think it will never betray them, but I do not agree

Polytoken 0.4.0-unstable.3 is up. Sonnet 5 now works. The Anthropic provider did not have a bug. Anthropic changed the way their API worked without telling anyone. They do that. Real rewarding to support. brew install polytoken/tap/polytoken-unstable docs.polytoken.dev/installation...

Polytoken 0.4.0-unstable.2 released. Adds /goal mode and the usual pile of fixes and improvements. Plugging away at the list. brew install polytoken/tap/polytoken-unstable `curl -fsS get.polytoken.dev/unstable | bash` docs.polytoken.dev/changelog/#0...

Polytoken 0.4.0-unstable.1 released on the new unstable channel, because this moves some internal, operational cheese and could use some testing. By the adventurous. brew install polytoken/tap/polytoken-unstable `curl -fsS get.polytoken.dev/unstable | bash` docs.polytoken.dev/changelog/#0...

It’s… what I do.

Polytoken 0.3.3 released! This is a small update with one significant fix to MCPs in subagents, because @sifu.tweety.fish broke everything. brew install polytoken/tap/polytoken docs.polytoken.dev/installation

Polytoken 0.3.2 released last night, but I forgot to post about it. I'm very smart. Custom headers for providers (makes Bedrock work), an (ahem) working 256-color mode, and the usual pile of bugs and tweaks. brew install polytoken/tap/polytoken docs.polytoken.dev/installation

Polytoken 0.3.1 is up to fix a bug whose existence is steadfastly denied by our changelog. You should update, though, for those zero bug fixes that definitely aren't recommended. brew install polytoken/tap/polytoken docs.polytoken.dev/installation

fyi - plan mode regression in 0.3.0. my local environment was hiding something (I need a more hermetic env). will have an 0.3.1 up shortly to fix.