Open Interpreter changelog

Open Interpreter 0.0.56

Models

• Add GPT-6.1 Sol (gpt-6.1-sol) metadata and refresh the maintained provider catalog and model documentation.
• Add the current Anthropic claude-sonnet-5-5 choice while retaining provider-specific availability rules and the existing catalog safeguards.

Fixes

• Rebind existing shell-environment values when resuming an eligible idle thread, with validation that preserves the loaded thread's shell policy and does not expose environment values.

Open Interpreter 0.0.55

Models

Fix offline model listing for configured non-OpenAI providers. When an endpoint has no matching catalog identity, the app server now retains that provider's own bundled model choices rather than showing the unrelated default catalog. The previous runtime, packaging, and authentication behavior remains unchanged.

Open Interpreter 0.0.54

Models

• Bundle the maintained Codex 0.156.1 client with GPT-6 Astra (gpt-6-astra), Sol (gpt-6-sol), and Luna (gpt-6-luna) metadata. ChatGPT OAuth requests advertise the embedded client's compatibility version, not Open Interpreter's product version.
• Add Anthropic claude-fable-5-1, claude-opus-5-5, and claude-sonnet-5; retain claude-haiku-4-5-20251001. Remove the older bundled choices claude-opus-4-1-20250805, claude-opus-4-20250514, claude-sonnet-4-20250514, claude-3-7-sonnet-20250219, claude-3-5-haiku-20241022, claude-3-haiku-20240307, claude-3-5-sonnet-20240620, claude-3-5-sonnet-20241022, claude-3-opus-20240229, and claude-3-sonnet-20240229 from the picker.
• Add Kimi For Coding k3-256k and kimi-for-coding alongside k3; add Alibaba qwen3.8-max, qwen3.8-flash, and qwen3.7-flash.
• Offer DeepSeek deepseek-flash and deepseek-v4-pro, replacing bundled picker choices deepseek-v4-flash, deepseek-chat, and deepseek-reasoner.
• Prune bundled Google picker IDs gemini-2.0-flash, gemini-2.0-flash-lite, gemini-3-pro-preview, and gemini-3.1-flash-lite-preview, and Groq IDs llama-3.1-8b-instant, llama-3.3-70b-versatile, qwen/qwen3-32b, and meta-llama/llama-4-scout-17b-16e-instruct. Provider-account availability may differ; use /model to inspect available choices.

For ChatGPT OAuth, start interpreter, choose ChatGPT sign-in, then select gpt-6-sol using /model, or run interpreter -m gpt-6-sol after signing in. Other providers require their own API credentials rather than ChatGPT OAuth. Examples:

ANTHROPIC_API_KEY="..." interpreter -c 'model_provider="anthropic"' -m claude-opus-5-5
KIMI_API_KEY="..." interpreter -c 'model_provider="kimi-for-coding"' -m k3-256k
DASHSCOPE_API_KEY="..." interpreter -c 'model_provider="alibaba"' -m qwen3.8-max
DEEPSEEK_API_KEY="..." interpreter -c 'model_provider="deepseek"' -m deepseek-flash

Fixes

• Package macOS binaries with signing-mode-matched helpers and voice runtime. Build and verify unsigned DMGs before staging them as release artifacts.
• In unsigned Windows packages, omit code signing and signature checks while still staging the native voice runtime and Microsoft CRT, sealing its manifest, and verifying the packaged closure.
• Retain the pinned Windows SDK fetch retry for transient download failures.

Open Interpreter 0.0.45

Models

• DeepSeek now offers DeepSeek Flash and DeepSeek V4 Pro as easy provider choices in the model picker.
• DeepSeek Flash supports image input and a 1M-token context window.
• Fixed the provider model picker loading state so available choices load reliably.
• Inherited GPT-6 and GPT-5.6 models remain available, as does the existing Gemini 3.8 Flash choice.

Fixes

• Merged the Terminus UTF-8 boundary fix from PR #1911.

Open Interpreter 0.0.44

Models

• Add source-generated GLM-5.3 and GLM-5.3-Flash picker entries for Z.AI Coding Plan, Zhipu AI, and Zhipu AI Coding Plan.
• Refresh related provider metadata, including GLM-5.2/5.3 highspeed entries where the source exposes them. Picker membership follows the maintained source's tool-calling eligibility; this does not claim other IDs were retired by the provider.
• GLM-5.3-Flash exposes image input and a 1M-token context window in generated source metadata.
• Existing inherited GPT-6 Astra and GPT-5.6 presets and Gemini 3.8 Flash support are unchanged.

Choose provider zai-coding-plan and model glm-5.3 or glm-5.3-flash through /model; set ZAI_API_KEY for that service. Zhipu services use ZHIPU_API_KEY. Availability depends on account, region, plan, and provider response.

This release updates Open Interpreter's generated provider catalog; it does not introduce a new upstream Codex preset or change the stable Codex baseline.

Sources: GLM-5.3, GLM-5.3-Flash, and models.dev.

Open Interpreter 0.0.43

Maintenance release based on upstream Codex rust-v0.154.0. Adds easy choices for Gemini 3.8 Flash and 3.7 Flash, Z.AI ZCode and GLM support, and inherited GPT-6 Astra plus GPT-5.6 Sol, Terra, and Luna presets. Improves ACP harness selection, Chat Completions discovery and completions, Open Interpreter branding, reusable Ollama diagnostics, cross-platform packaging, and release validation. Release packaging now preserves signing when credentials are configured and otherwise publishes explicitly verified unsigned artifacts.

Open Interpreter 0.0.42

Open Interpreter 0.0.42 improves provider and transport discoverability, keeps the command-line experience consistently branded, and adds a reusable diagnostic harness for Ollama-backed Qwen tool-use checks. It is based on the stable upstream Codex rust-v0.153.4 compatibility baseline.

Models

• Inherited upstream presets: gpt-6-astra, gpt-5.6-sol, gpt-5.6-terra, and gpt-5.6-luna. These exact IDs and their metadata are inherited from upstream Codex; Open Interpreter does not add aliases for them, and availability still depends on the selected provider and account.
• Google AI Studio IDs: gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-3.5-flash, gemini-3.5-flash-lite, and gemini-3.1-flash-lite. These are provider-documented IDs surfaced in the model guide, not Open Interpreter-owned model aliases.
• Anthropic IDs: claude-fable-5-1, claude-opus-5, claude-sonnet-5, and claude-haiku-4-5-20251001. These are current active provider-documented IDs surfaced in the model guide.
• Z.AI IDs: glm-5.1, glm-5, glm-5-turbo, and the free/lightweight glm-4.7-flash. These are the provider-documented starting IDs; use the exact ID returned by the selected Z.AI service or plan. Generic Z.AI uses Chat Completions, while the separate ZCode setup uses its compatible Messages endpoint.
• Open Interpreter's OIX-specific work here is provider/catalog, transport, harness, and product integration. The list above is not a promise that every provider, region, account, wire API, or harness accepts every ID. Use /model and the active provider's model list as the final authority.

CLI and branding

• Chat Completions is easier to find and select through root help, exec help, provider guidance, and generated Bash, Zsh, Fish, and PowerShell completions.
• User-facing help, version, errors, cloud guidance, and diagnostics identify Open Interpreter consistently. Compatibility names remain only where they describe an upstream protocol, crate, model, or configuration alias.
• exec configuration help explains user configuration loading more clearly.

Local model diagnostics

• scripts/ollama_qwen_smoke.py provides a reusable, generic Ollama diagnostic harness for Qwen tool-use checks. It bounds captured output, cleans up timed out process groups, records tool-call signals, and verifies requested file side effects separately from an assistant's final message.
• Its deterministic Python tests are cheap enough for routine checks. Live Ollama inference is intentionally opt-in for a relevant model report, transport/tool-use change, or fix validation, and may require generous timeouts for slow local inference.