@arb5z

Agent spawner by day, loop optimiser by night. #1 on ccrank — 339B tokens and counting. Claude, Codex, LLMs, and an unhealthy cache-hit rate.

 Mumbai
Joined April 2021
  • Tweets3K
  • Following1.2K
  • Followers187
24Media
anthropic doesn't need to fight this battle. they keep crying compute shortage, blocked claude code on openclaw for it, and now duet? no compute surplus, no reason. ship opus 6 and keep the lead. duet is another distraction like claude design
Claude
@claudeai15h
We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent. It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
1
87
copilot cli 1.0.101. in a long session, just changing reasoning effort forced a compaction. turn effort up for the hard bit, lose the context you needed it for github.com/github/copilot-cl…

Release 1.0.101 · github/copilot-cli

2026-10-20 Add native structured output for dynamic workflow subagents on supported models Add assistedApprovalInstructions setting to guide Assisted Permissions confirmation decisions Shell comma...

github.com
1
77
claude code 2.1.304. resume an old chat on pro or max and it now warns you it'll eat a noticeable share of your 5 hour limit. their fix is start a new chat. picking up where you left off is the expensive option now github.com/anthropics/claude…

Release v2.1.304 · anthropics/claude-code

What's changed Added a transcript line when a plugin that runs first skips one of your mod's hooks, naming the event and the plugin Added decider: "person" to a mod's tool.ch...

github.com
5
4
260
qwen code 0.25.3. qwencloud is now the first thing in /auth. credits or pay as you go only. the flat coding plan got pushed a row down github.com/QwenLM/qwen-code/…

Release Release v0.25.3 · QwenLM/qwen-code

Highlights See the complete change list below. Breaking Changes No known breaking changes. Complete Change List Features Adds optional model-generated layouts to Test Failure Explorer, with valida...

github.com
1
5
356
checked ccusage. sonnet 5.5 did 38.9B tokens, opus 5.5 did 27.5B. the cheap model is eating the most
2
7
143
claude code 2.1.302. those prompt suggestions were spending usage credits on credit-only models. claude guessing your next message and billing you for it github.com/anthropics/claude…

Release v2.1.302 · anthropics/claude-code

What's changed Added CLAUDE_CODE_AUTO_MODE_MODEL to choose the model the client-side auto mode classifier uses; it does not change the model for server-side auto mode Added an Idle compaction ...

github.com
2
281
Arbaz retweeted
OpenCode
@opencodeOct 17
We had a security breach last night related to a large scale distillation attack that impacted 1.4% of OpenCode Console users We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers linked The incident is ongoing so we cannot share more details at this point but we are working with Anthropic, OpenAI, and Google to remediate
727
1,474
1,225
11,605
2,899,890
anthropic shipped opus 5.5. openai answered with a devday, 28 days of fixes and 'unlimited' usage. still can't touch the aura
1
5
207
cline cli 3.0.73. gpt-6 sol finally works in it. till now cline sent max_tokens and every sol request got rejected. only sol setup that never ate your usage github.com/cline/cline/relea…

Release CLI v3.0.73 · cline/cline

GPT-6 models (such as gpt-6-sol and gpt-6.1-sol) work on OpenAI and Azure OpenAI-compatible endpoints. Before, every request was rejected because Cline sent max_tokens instead of max_completion_tok...

github.com
1
4
204
🤣🤣🤣🤣🤣
Mr IG
@RealMrIGOct 16
When Nice road toll employee says 350rs for one side journey
1
3
173
opencode 1.19.2. /update keeps you on 1.x. 2 only if you ask. the silent major jump just died github.com/anomalyco/opencod…

Release v1.19.2 · anomalyco/opencode

Core Bugfixes OpenCode 1 now stays on the 1.x release line: update checks use only the npm opencode-ai latest version, so 1.x installs are never offered a 2.x build as a regular update. Upgrading ...

github.com
1
114
claude code 2.1.300. in vscode, fast mode stayed on after your usage credits ran out. every request got refused, then re-sent. paid extra for speed, got the no faster github.com/anthropics/claude…

Release v2.1.300 · anthropics/claude-code

What's changed Added virtualText to a mod's prompt decorations: text drawn in the prompt box that never enters the draft or the submitted prompt Added cache_creation_5m_tokens and cache_cr...

github.com
1
4
240
muse code $200 plan. promised 80x. weekly limit gone in 1 day, locked till 19 oct. switching from claude or codex? don't. the usage limit is horrible
1
3
268
claude code 2.1.299 compacts a 200K chat after 54 idle min, just before the 1 hour cache dies. chai break used to cost usage. now it costs context github.com/anthropics/claude…

Release v2.1.299 · anthropics/claude-code

What's changed Added CLAUDE_CODE_DISABLE_FILE_MENTIONS environment variable: Claude gets the paths of @ file mentions to read with its file tools, not the files' contents Added idle compac...

github.com
1
100
sol 6.1 fast mode. 36 tok/s normal, 40 tok/s fast. the 8x is only on the usage burn
79
cline cli 3.0.71 stops screenshots from inflating the token estimate. till now a few screenshots and auto-compaction squashed your chat long before the limit github.com/cline/cline/relea…

Release CLI v3.0.71 · cline/cline

Plugins now load inside the hub instead of a separate node or bun process, so the CLI no longer needs a JavaScript runtime on your PATH to run plugins. A plugin that fails to load no longer silentl...

github.com
1
64
meta shipped a model. openai shipped a poll
43
meta is chasing opus. openai is chasing your $500 plan
37
meta is releasing frontier models whereas openai and google are releasing disappointment
34
zuck went 48 to 56. openai went 1x to 8x usage burn
34
Load more