[AINews] Pi 1.0, Pi Durable, and AIE NYC

Earendil released Pi 1.0 and Pi Durable, both of which reached the front page of Hacker News, with Pi Durable porting Pi to TypeScript and externalizing all stateful components so agents resume from their last exact checkpoint after a crash. Pi 1.0 adds native MCP support via Codemode, extension support for virtual models, deferred tool loading, cache warming for Anthropic models, mid-conversation system messages, a new TUI theme, and full-screen mode by default. Pi Durable runs on any JavaScript runtime including Node, Bun, and Cloudflare with pluggable Memory, SQLite, and JSONL storage backends, and supports parallel branching conversations, background context compaction, multiplayer state sync, and hot-swapping tool code while the agent runs.

Last call for regular tickets for AI Engineer NYC https://ai.engineer/nyc See you in 2 weeks As an exclusive for Latent Space subscribers, the first 30 of you can take a 30% off code https://app.ai.engineer/e/ai-engineer-new-york-2026?discount=LS-NYC26 if it helps for new tickets only, no refunds . Pi is often mentioned in the same breath as OpenClaw, as we did earlier this year https://www.latent.space/p/ainews-sci-fi-with-a-touch-of-madness?utm source=publication-search : but today is time for the increasingly well regarded Earendil https://www.youtube.com/watch?v= Zcw sVF6hU&pp=0gcJCS4MAYcqIYzv , which Pi joined, to have its day in the sun, with both Pi 1.0 https://news.ycombinator.com/item?id=49926069 and Pi Durable https://news.ycombinator.com/item?id=49925969 hitting the front page of HN. - Codemode https://earendil.com/posts/you-said-no-mcp/ native support for MCP, Jev and image models - Extension support for virtual models https://github.com/earendil-works/pi/releases - Deferred tool loading - Cache warming for anthropic models - Mid-conversation system messages https://github.com/earendil-works/pi/blob/main/packages/coding-agent/docs/session-format.md transcript-aware prompt and tool changes - A new TUI theme - Full-screen mode by default Pi Durable https://earendil.com/posts/pi-durable/ ports Pi to TypeScript and externalizes all stateful components of Pi: - Crash Survival: Every step is recorded as a checkpointed task. If a process fails or restarts, agents and subagents automatically resume from their last exact state. - Portability: It runs anywhere with a JavaScript runtime like Node, Bun, or Cloudflare and uses pluggable storage backends Memory, SQLite, JSONL and flexible remote or local execution environments. - Concurrency: A single harness can run multiple parallel, branching conversations—such as a main channel and separate threads—without blocking one another. - Extensibility: Developers can bundle custom system prompts, tools, hooks, and durable tasks e.g., multi-step checkout processes with rollback capabilities into installable “Extensions.” - Context Management: Automatic background compaction summarizes older messages to maintain token limits without pausing the agent’s active work. - Multiplayer & State Sync: Application state like a to-do list is stored in documents directly alongside the conversation transcripts, allowing multiple users or UIs to connect, watch, and steer the same agent simultaneously. - Hot-Swapping: Tool and extension code can be updated dynamically while the agent is running, with the next tool call automatically picking up the new code. AI News for 10/01/2026-9/30/2026. We checked 12 subreddits, 544 Twitters https://twitter.com/i/lists/1585430245762441216 and no further Discords. AINews’ website https://news.smol.ai/ lets you search all past issues. As a reminder, AINews is now a section of Latent Space https://www.latent.space/p/2026 . You can opt in/out https://support.substack.com/hc/en-us/articles/8914938285204-How-do-I-subscribe-to-or-unsubscribe-from-a-section-on-Substack of email frequencies AI Twitter Recap Frontier and Multimodal Launches: Gemini 4 Argon, GPT-6.1 Sol and FLUX 3 - Gemini 4 Argon : Google announced a new generation of Gemini, with contributors highlighting revised pretraining mixtures, long-horizon post-training data, and internal applications in memory optimization, code migration and mathematics. These are developer accounts of how the model was built and used—not independent evidence of general superiority Google researcher https://x.com/mirrokni/status/2105500370675921213 . - Validation : Google says new Gemini revisions now undergo weeks of testing by thousands of internal software engineers before release Logan Kilpatrick https://x.com/OfficialLoganK/status/2105521401566486875 . - Contested readiness : A circulated Bloomberg report attributed coding weaknesses to anonymous insiders; a subsequent post reported a senior DeepMind engineer rejecting that account. Treat the practical coding-quality dispute as unresolved, rather than interpreting either benchmarks or employee reactions as decisive reported criticism https://x.com/kimmonismus/status/2105570914574209283 , reported rebuttal https://x.com/kimmonismus/status/2105709302484729879 . - GPT-6.1 Sol : OpenAI’s update is primarily an efficiency story. Sam Altman called it the company’s fastest-growing model and said serving performance had improved after launch-time load problems update https://x.com/sama/status/2105688354834756036 . - Measured economics : Artificial Analysis reports $0.72 per Intelligence Index task at maximum effort, versus $1.04 for GPT-6 Sol and $3.26 for Astra. Fewer turns and cheaper cache reads—not simply fewer generated tokens—drive the improvement results https://x.com/ArtificialAnlys/status/2105491868608004578 , explanation https://x.com/ArtificialAnlys/status/2105449959554441580 . - Multimodal fix : OpenAI also corrected image encoding for Luna and Sol. Luna gained one Intelligence Index point, including improvements on visual-document and knowledge-work evaluations; Sol changed negligibly measurement https://x.com/ArtificialAnlys/status/2105491868608004578 . - Solar Mini 4 : Upstage’s proprietary text-only reasoning model reports 35B total/3B active parameters, a 1M-token context window and 262K maximum output. Weights are not released, so parameter counts remain vendor-reported analysis https://x.com/ArtificialAnlys/status/2105459219059401036 . - Pricing : $0.10/$0.40/$0.01 per million input/output/cache-hit tokens. - Trade-offs : Artificial Analysis scores it 24 overall and 83% on long-context reasoning, but only 1% on Terminal-Bench 4.0. Despite 208 tokens/s output, approximately 88K output tokens per task produce a 7.1-minute average completion time and roughly five times Luna’s task cost. - FLUX 3 Image : Black Forest Labs launched native generation up to 4K, up to ten reference images, bounding-box layout control and targeted multi-turn editing. Preserving every untouched pixel is a vendor capability claim, not independently established here announcement https://x.com/bfl ai/status/2105734605621825738 . - Interactive video agents : Tavus introduced Griffin, a video-to-video interaction model. It claims 48% of live participants mistook it for a human, versus under 3% for earlier systems; that result should not be generalized into an unrestricted “Turing test passed” conclusion without the test protocol announcement https://x.com/tavus/status/2105704169009246248 . - Enterprise deployment : Separately, Synthesia launched Sessions: conversational avatars for roleplay and survey interviews, extending its previous one-way training-video product launch https://x.com/synthesiaIO/status/2105591908659540474 .