ByteBrief Tech Insights

we shipped a new version of gpt-5.5 instant today. the previous model was too bullet pilled. the new one improves on some other important dimensions: sycophancy, factuality, and multilingual performance. hope you'll like it! always interested in feedback

test confirms it: :danger-no-sandbox is rejected by 0.135.0, while :danger-full-access works and resolves to unrestricted filesystem plus enabled network. The remaining question is why the TUI is choosing the old name; the likely answer is a stale running session or a 0.135.0 TUI persistence bug.

@thsottiaux CLI update 0.135.0 BUG: "Failed to set permission profile `:danger-no-sandbox`: Failed to rebuild config for permission profile :danger-no-sandbox" What is this Quality Assurance before release? jesus.... any help?

test confirms it: :danger-no-sandbox is rejected by 0.135.0, while :danger-full-access works and resolves to unrestricted filesystem plus enabled network. The remaining question is why the TUI is choosing the old name; the likely answer is a stale running session or a 0.135.0 TUI persistence bug.

@CodexReleases CLI update 0.135.0 BUG: "Failed to set permission profile `:danger-no-sandbox`: Failed to rebuild config for permission profile :danger-no-sandbox" What is this Quality Assurance before release? jesus.... any help?

Codex CLI 0.135.0 is out. Highlights: - `codex doctor` now reports richer environment, Git, terminal, app-server, and thread inventory diagnostics - Vim mode gained text-object editing, improved word/line-end behavior, and a configurable interrupt-turn binding - `/status` shows remote connection details and server version when connected over a remote transport - Python SDK now exposes friendly Sandbox presets for thread and turn APIs Complete details in thread ↓

test confirms it: :danger-no-sandbox is rejected by 0.135.0, while :danger-full-access works and resolves to unrestricted filesystem plus enabled network. The remaining question is why the TUI is choosing the old name; the likely answer is a stale running session or a 0.135.0 TUI persistence bug.

@fcoury CLI update 0.135.0 BUG: "Failed to set permission profile `:danger-no-sandbox`: Failed to rebuild config for permission profile :danger-no-sandbox" What is this Quality Assurance before release? jesus.... any help?

Codex CLI v0.135.0 is out with the new cleaner markdown table styling I teased before 👇 🧵

GPT 5.6 DROPPING SOONNNN 🔥🤖💥🚀 Mind-blowing upgrades incoming, world-changing vibes only!!! 🎉🌟🧠✨😍💯

@sandislonjsak You'll be back after one prompt if I can trust the timeline

Claude Opus 4.8 Fast just created the best Remotion video I have ever seen. One shot. No revisions. A complete BridgeSpace marketing video generated from a single prompt. Grok 4.3 took 25 minutes just to plan a Remotion video and the output was garbage. Claude Opus 4.8 one shotted a better version instantly. This model is on a completely different level.

introducing thermos in cursor a deep security/correctness audit and a harsh code quality audit, run in parallel on your branch, synthesized into one prioritized list

@ClaudeDevs @grok summarize what is ultra code from antrhopic on claude opus 4.8. It just dropped. How is it different from xhigh?

New in Claude Code (research preview): dynamic workflows. Claude writes an orchestration script on the fly, then spins up a large fleet of coordinated subagents in parallel to take on your most complex tasks. Use the word "workflow" in a prompt to get started.

This guy @bridgebench just opened up 30 terminals with Claude Opus 4.8 and just one of them in one shot vibe-coded a browser shooting game. WTF is 2026 🤯🤯🤯 @bridgebench @AnthropicAI

Claude Opus 4.8 is now available in Cursor. On CursorBench, it's able to work much more efficiently than Opus 4.7. We've also found it to be more persistent on harder tasks.

Opus 4.8 is being prepared for release today by @AnthropicAI 🔥 We might witness a rare dual release by OpenAI and Anthropic. Are you READY?

Hit GitHub's rate limit one too many times, so I built octopool: a Cloudflare Worker that pools your team's PATs + GitHub App installations behind a shared read cache. Self-host on Cloudflare. Drop-in gh shim. octopool.dev

@elonmusk Holy shit

SpaceX has almost finished writing V1.0 of an in-house AI training stack in C that exact-maps to 220k GB300s with 800G NICs, making heavy use of pipeline parallelism and getting as close to bare metal as possible. The potential speed improvement vs JAX for large training runs is over an order of magnitude.

Hey @romainhuet I cannot with CODEX CLI BUG on LINUX... -> OUTPUT FROM CLI: "apply_patch is failing on existing files because the filesystem sandbox cannot read through bwrap in this session, although it created the new files earlier. I’m going to use narrowly scoped scripted edits with approval because the remaining work is documentation consolidation, not code behavior." Someone help me out.. In my config: default_permissions = ":workspace" approvals_reviewer = "guardian_subagent" ...any help? Im on latest version of codex

@thsottiaux OR ANYONE me out with this bug.. im on the latest codex cli version on LINUX: -> OUTPUT FROM CLI: "apply_patch is failing on existing files because the filesystem sandbox cannot read through bwrap in this session, although it created the new files earlier. I’m going to use narrowly scoped scripted edits with approval because the remaining work is documentation consolidation, not code behavior." Someone help me out.. In my config: default_permissions = ":workspace" approvals_reviewer = "guardian_subagent"

Team is investigating elevated latencies across the board for API and ChatGPT. Updates will be posted on status.openai.com