Skip to content
Agents tracked: 492 Releases (7d): 545 Agent status: 1 with issuesSince last limit reset: Claude 9d · Codex 7d For AI agents: Markdown Updated Nov 20, 2026

Open Interpreter changelog: what's new each month

Every stable Open Interpreter release summarised by month: the highlights, new features, improvements, fixes and anything you need to act on. 3 months covered; the current month updates daily.

November 2026 · so far

4 releases: 0.0.58 → 0.0.61

November brought stability improvements and internal enhancements to Open Interpreter. The month focused on compatibility validation, durability features, and build integrity.

Highlights

  • Added owner-bound, read-only background terminal status tracking.
  • Implemented durable creation IDs that persist across response loss and restarts.
  • Validated compatibility with Codex 0.159.0 profile.
  • Enhanced Linux archive attestation and checksum verification.

New

  • Owner-bound, read-only background terminal status with bounded terminal access.
  • File-backed managed Goal creation contract.
  • Durable creation IDs across response loss and restart scenarios.
  • Linux Open Interpreter archive attestation and checksum manifest.
  • Build provenance verification before artifact publishing.

Improved

  • Legacy feedback delivery now works alongside optional signed attachments.
  • Creation ID durability with atomic operations.

October 2026

2 releases: 0.0.56 → 0.0.57

October brought model updates and improved session handling for Open Interpreter. The month focused on maintaining compatibility with the latest AI providers and refining the interpreter's ability to resume previous work.

Highlights

  • Added support for GPT-6.1 Sol model
  • Updated to Anthropic's Claude Sonnet 5.5
  • Improved shell environment preservation when resuming idle sessions
  • Validated compatibility with Codex 0.159.0

New

  • GPT-6.1 Sol model support

Improved

  • Shell environment values are now rebound when resuming eligible idle sessions
  • Anthropic provider refreshed with current Claude Sonnet 5.5 option
  • Maintained provider metadata updated

September 2026

6 releases: 0.0.42 → 0.0.55

September brought expanded model support across multiple AI providers and refinements to model selection and packaging. The month focused on surfacing current provider-documented model IDs and improving the picker experience.

Action needed
  • Removed older Anthropic bundled choices: claude-opus-4-1-20250805 and claude-opus-4-20250514.
  • Removed older Google picker IDs: gemini-2.0-flash, gemini-2.0-flash-lite, gemini-3-pro-preview, and gemini-3.1-flash-lite-preview.
  • Removed Groq bundled IDs including llama-3.1-8b-instant and llama-3.3-70b-versatile.
  • DeepSeek picker choices replaced: deepseek-v4-flash, deepseek-chat, and deepseek-reasoner replaced by deepseek-flash and deepseek-v4-pro.

Highlights

  • Added support for latest models from OpenAI (GPT-6 Astra, Sol, Luna), Anthropic (Claude Fable, Opus 5, Sonnet 5), Google (Gemini 3.8 and 3.7 Flash), and others.
  • DeepSeek Flash and V4 Pro now available with image input and 1M-token context window support.
  • Improved model picker loading and selection through enhanced help documentation and clearer configuration guidance.
  • Added GLM-5.3 and GLM-5.3-Flash from Z.AI with image input and 1M-token context support.
  • Improved macOS and Windows package signing and verification processes.

New

  • Support for Kimi For Coding k3-256k model alongside existing k3.
  • Alibaba Qwen models: qwen3.8-max, qwen3.8-flash, and qwen3.7-flash.
  • GLM-5.3 and GLM-5.3-Flash pickers for Z.AI Coding Plan and Zhipu AI.
  • Enhanced offline model listing that retains bundled choices when endpoints have no matching catalog identity.

Improved

  • Chat Completions is easier to find and select through root help and exec help.
  • User-facing help, version messages, errors, and diagnostics now identify configuration more clearly.
  • Exec configuration help explains user configuration loading with greater clarity.
  • Provider model picker loading state now shows available choices properly.
  • Maintained Codex client bundled with current GPT model metadata.
  • macOS binaries now packaged with signing-mode-matched helpers and voice runtime.
  • Windows packages include native voice runtime and Microsoft CRT with verified manifest.

Fixed

  • Fixed provider model picker loading state so choices display correctly.
  • Merged Terminus UTF-8 boundary fix.
  • Fixed offline model listing for configured non-OpenAI providers.
  • Improved Windows SDK fetch retry handling for transient download failures.

Summaries are written automatically from the official release notes (full changelog ↗); check the original notes before relying on a detail. Open Interpreter: pricing, features and alternatives · All changelogs