Developer 3s Key Decision Metrics
On 2026-10-01 Earendil shipped Pi 1.0 (MIT): a hardened minimal coding-agent harness. Built-ins: Codemode (QuickJS sandbox for model-written JS calling tools / Jev / image models) + MCP, virtual-model routing, deferred tools, Anthropic cache warming, mid-conversation system edits. At ingest: 111793★; HN 1645 pts / 573 comments. Experimental Pi Durable shipped the same day.
Key Takeaways
- ✓Official post 2026-10-01; GitHub v1.0.0 at 2026-10-01T19:20:55Z; MIT
- ✓At ingest: 111793★; HN Pi 1.0 thread 1645 pts / 573 comments
- ✓1.0: Codemode+MCP, virtual models, deferred tools, Anthropic cache warming, mid-convo system messages, fullscreen TUI; changelog ~5300→3300 prompt tokens (−40%) for GPT-5.6 with defaults+codemode
- ✓Not NVIDIA SoL-Pi (already in D1) — this is Earendil’s Pi harness
- ✓Install: curl https://pi.dev/install.sh | sh or npm @earendil-works/pi-coding-agent; docs pi.dev
Turn your technical choice into a development budget
Compare 29+ dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Core Background & Industry Pain Points
Coding agents keep shipping kitchen-sink defaults. Earendil’s bet is the opposite: a minimal, supermalleable harness—extend via packages/skills instead of baking every workflow into core. On 2026-10-01 they shipped Pi 1.0 (stable) plus experimental Pi Durable. Repo earendil-works/pi: 111793★ at ingest; HN thread 1645 pts / 573 comments. Distinct from NVIDIA SoL-Pi already in D1.
Architecture & Mechanisms
1.0 highlights: Codemode (QuickJS sandbox for model-written JS calling tools / Jev / image models), native MCP, virtual models (e.g. Opus plan → Jev switch → GPT-6 Luna implement), deferred tools, Anthropic cache warming, mid-conversation system edits, fullscreen TUI by default. Modes: interactive, print/JSON, RPC, SDK.
Benchmarks & Measured Numbers
No official SWE-bench/Terminal-Bench numbers in the launch post—none invented here. Verified: v1.0.0 at 2026-10-01T19:20:55Z, MIT; changelog example ~5300→3300 prompt tokens (−40%) for GPT-5.6 with defaults+codemode; GitHub 111793★; HN 1645/573.
Developer Quickstart
curl -fsSL https://pi.dev/install.sh | sh or npm @earendil-works/pi-coding-agent. Enable "defaultTools": ["+codemode"]; configure mcp.json; try jev-router virtual model. Docs: pi.dev.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.