We're also open-sourcing our memory format: - support graph relationships - updates over time with historical records - backed by git and markdown and you can use it with any agent, not just Devin!

Did Opus 5.5 get nerfed or improved? I asked for a stained-glass hibiscus on Sept 30 and again today. Exact same prompt. Look at the difference in those petals.

Introducing Griffin, the first model to pass the video Turing test. 48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video. It’s the first Human Interaction Model (HIM).

is AI making us dumber or smarter

sorry gpt 6 astra lovers it might be the best model on some tasks but it has no personality, and utterly boring would avoid for ANY creative work

We are going to need a different AGI benchmark. Where is the goalpost moving next?

I will pay an absurd amount of money to anyone who builds a desktop app that lets me see all my coding agents across claude code, codex and cursor all at once Why hasn't anyone built this already

The banked reset has landed, I repeat, the banked reset has landed. Have an amazing weekend.

Announcing our new Speech Agent Arena, evaluating Speech to Speech models on real-world scenarios to analyze conversational preference and task success rate Existing Speech to Speech benchmarks cover reasoning, simulated agentic tasks, and conversational dynamics such as turn-taking and interruption handling. The Speech Agent Arena compares models and cascaded systems as humans complete real-world tasks, measuring conversational preference and successful tool use. This allows us to provide an evaluation which closer reflects real-world use, offering insight into which models users most prefer speaking with and how effectively those models support their requests. Overview of the Speech Agent Arena and Task Success Rate Human participants compare two hidden Speech to Speech models on the same assigned scenario, one of 15 agentic scenarios (tasks requiring tool calling, such as ordering takeout) or 20 non-agentic scenarios (tasks without tool calling, such as asking about opening hours). After separate live conversations with each model, participants select which they preferred, with these pairwise votes used to fit a Preference Elo score. For agentic scenarios, Task Success Rate is the share of eligible conversations (no participant deviations or unverifiable cases) where the model completed the requested action through the correct final tool call or calls. For all but a New Patient Dental Booking example, scenario model prompts, tool schemas and participant instructions are currently private to reduce overfitting. Additionally, the Speech Agent Arena currently uses a qualified pool of paid, screened third-party participants to conduct and evaluate agent interactions. Key results: ➤ Arena Preference Elo: @GoogleAI Gemini 3.1 Flash Live Preview - Minimal leads at 1,046 Elo, followed by Gemini 3.1 Flash Live Preview - High at 1,014, @OpenAI GPT-Realtime-1.5 at 1,000, GPT Realtime (Aug '25) at 944, and @ElevenLabs Agents (default cascaded system of Scribe v2 Realtime / GPT-4o Mini / Eleven v3, with pre-registered tool schema) at 937. In reviewed conversations, highly preferred models tended to respond quickly, sound more natural and produce fewer unnatural sounds or audio artifacts ➤ Task Success Rate: @spacexai Grok Voice Think Fast 2.0 High leads at 94.7%, followed by @OpenAI GPT-Realtime-2.1 High at 91.5%, @ElevenLabs Agents (Default Cascaded System) at 90.5%, and GPT-Realtime-2 (High) at 89.8%, with GPT Realtime (Aug '25) and GPT-Realtime-2.1 Minimal tied at 89.4%. Gemini 3.1 Flash Live Preview - Minimal leads overall preference at 1,046 Elo but records a 74.6% Task Success Rate, showing that a preferred conversation does not always result in successful task completion - some conversations can sound as though the requested action was completed even when the required final tool call was unsuccessful We are continuing to expand our coverage of native and cascaded Speech to Speech systems, and welcome feedback as we add more models, providers and scenarios. See more details below ⬇️

We just released Qwen3.8-27B MLX quants. 🍎🐳 This time: the official Qwen3.8-27B, packaged for Apple Silicon + MLX. 4-bit · 6-bit · 8-bit Up to 98.34% Top-1 agreement vs BF16. Native MLX weights. No GGUF conversion required. Fast. Private. Local. Have fun! → huggingface.co/orcarouter/Qw…

Fable 5 Ultracode for everything or bust

For those wondering which to care about: Excellent: - Codex (good @ everything) - Devin (best eng harness) - Claude Code (access Fable) Pretty darn good: - Pi (Oh My Pi unless you have time) - Cursor (but do you trust Elon) Not worth: - T3 Code - Grok Bot - OpenCode / Desktop

It's me again. I come bearing great news. First of all, we have hit 20M active users for Codex some time this week. Second of all, this is cause for celebration and during the day we will credit every Codex and ChatGPT Work user with a BANKED reset that you can use at your own leisure. And we will have some other good news later too! Now, on usage limits draining faster, while we're not seeing anything abnormal, we do take it incredibly seriously and there is an ongoing investigation. I will share if we do find anything and my below post is really a clarification on a specific pattern that we did see that I wanted to call out. Go do something amazing today.