Coding agent model support
Which models each coding agent runs on
We checked 30 coding agents against their makers' own pages. 22 run on Claude, 24 run on GPT, 17 run open-weight models and 19 take your own key.
| Coding agents | Claude | GPT | Open models | BYOK |
|---|---|---|---|---|
| AiderChecked | Yes | Yes | Yes | Yes |
| Augment CodeChecked | Yes | Yes | Partly | Partly |
| BoltChecked | Not checked | Not checked | No | Partly |
| Claude CodeChecked | Yes | No | No | Yes |
| ClineChecked | Yes | Yes | Yes | Yes |
| CodeRabbitChecked | Yes | Yes | No | Partly |
| Codex CLIChecked | No | Yes | Yes | Yes |
| CursorChecked | Yes | Yes | Partly | Yes |
| DevinChecked | Yes | Yes | Yes | Partly |
| Devin DesktopChecked | Partly | Partly | Partly | No |
| FactoryChecked | Yes | Yes | Yes | Yes |
| Gemini Code AssistChecked | No | No | No | Partly |
| GitHub CopilotChecked | Yes | Yes | Yes | Yes |
| Google Antigravity CLIChecked | Partly | Partly | Partly | No |
| gooseChecked | Yes | Yes | Yes | Yes |
| JugglerChecked | Yes | Yes | Yes | Yes |
| JunieChecked | Yes | Yes | Yes | Yes |
| Kilo CodeChecked | Yes | Yes | Yes | Yes |
| LettaChecked | Yes | Yes | Yes | Yes |
| ModurianChecked | Not checked | Yes | Not checked | No |
| Muse CodeChecked | No | No | No | Partly |
| OpenCodeChecked | Yes | Yes | Yes | Yes |
| OpenHandsChecked | Yes | Yes | Yes | Yes |
| OrbiChecked | Partly | Yes | Yes | Yes |
| Replit AgentChecked | Yes | Yes | Partly | Partly |
| WarpChecked | Yes | Yes | Yes | Yes |
| WTAgentChecked | Yes | Yes | Partly | Partly |
| ZedChecked | Yes | Yes | Yes | Yes |
| ZencoderChecked | Yes | Yes | Partly | Yes |
| zotChecked | Yes | Yes | Yes | Yes |
"Checked" is the oldest answer in the row.
- Claude
- Runs on Anthropic's Claude models.
- GPT
- Runs on OpenAI's GPT models.
- Open models
- Runs open-weight models, hosted by the maker or on your own machine.
- BYOK
- Takes your own API key, inference or cloud account.
The "Partly" answers
- Augment Code, Open models: Open-weight families appear as vendor-hosted choices through inference providers, but no page documents pointing the product at weights you host yourself. Source,
- Augment Code, BYOK: You can supply your own compute via daemons and encryption keys via CMEK, but inference is resold at provider list price plus a fee, not run on your own model keys. Source,
- Bolt, BYOK: Bring-your-own-cloud is an Enterprise deployment option; no tier lets you supply your own model API keys or inference provider. Source,
- CodeRabbit, BYOK: Bring-your-own inference exists only in the Enterprise self-hosted deployment, where the agent sends prompts to your own LLM account; hosted plans use CodeRabbit's providers. Source,
- Cursor, Open models: All inference is hosted: the catalogue includes GLM 5.2 from Z.ai and Kimi K3 from Moonshot, but there is no path to point Cursor at open weights you run. Source,
- Devin, BYOK: Your own AWS KMS key encrypts data at rest, but only on the Enterprise Assured deployment; no bring-your-own-inference is documented. Source,
- Devin Desktop, Claude: Claude is a premium model: the plan table marks premium models absent on Free and present from Pro upward; the model table lists Opus, Sonnet and Haiku variants. Source,
- Devin Desktop, GPT: Same gate as Claude: OpenAI frontier models come with Pro and above, while Free is marked "Limited model availability" with premium models crossed out. Source,
- Devin Desktop, Open models: Open-weight models are hosted by Cognition and included from Pro; the model cost table carries DeepSeek, Moonshot (Kimi), Z.ai (GLM) and Nvidia entries. You cannot point it at your own weights. Source,
- Gemini Code Assist, BYOK: Your key is an overflow path, not the billing model: once the daily agent quota is spent you can set geminicodeassist.geminiApiKey to a Gemini or Vertex AI key. Inference stays on Google. Source,
- Google Antigravity CLI, Claude: Claude Opus and Sonnet 5.5 are selectable, but only on paid Google AI Pro, Ultra or Enterprise; the free tier's Claude 4.6 entries are flagged for removal. Source,
- Google Antigravity CLI, GPT: No OpenAI GPT model is offered; the only GPT-named entry is the open-weight GPT-OSS-120b, which the table marks for removal and excludes from Enterprise. Source,
- Google Antigravity CLI, Open models: One open-weight model, GPT-OSS-120b, is selectable on consumer plans, and the same table dates its removal, leaving no standing open-weight option. Source,
- Muse Code, BYOK: You supply your own Meta Model API key, including in CI from a secret store, but inference still runs on Meta's hosted endpoint and no third-party key or own-cloud option is documented. Source,
- Orbi, Claude: The vendor states Orbi can wrap Claude Code as the coding engine inside its delivery line, but no Anthropic entry appears in the published provider template catalogue, so it is a claim without a configuration recipe. Source,
- Replit Agent, Open models: Open-source model providers are a real but second-class category for Agent: they exist as a provider policy and are off unless an admin turns them on. Source,
- Replit Agent, BYOK: You can supply provider keys for models your own app calls, but the inference behind Agent itself is fixed to Replit's infrastructure. Source,
- WTAgent, Open models: Providers include DeepSeek, Kimi and GLM, whose weights are published openly, but WTAgent drives their hosted web chat rather than running any weights locally. Source,
- WTAgent, BYOK: You bring your own inference, but as a signed-in web chat account in a dedicated Chrome profile rather than an API key or your own cloud. Source,
- Zencoder, Open models: Open-weight models are not sold in the managed catalogue; you reach them by declaring your own Ollama or vLLM endpoint in settings.json and paying for that inference yourself. Source,
Use the data
How we check listings: methodology.