--- title: "Best Models for OpenClaw (July 2026): Tested & Ranked" description: "Which AI model should you use with OpenClaw? We tested 50+ models. Claude Opus 4.8 for coding, Gemini 3.1 Pro for research, MiniMax M3 for budget. Updated regularly." date: 2026-01-24 updatedDate: 2026-07-23 location: San Francisco, CA – Jan 24th, 2026 image: /images/best-models-for-clawdbot-hero.jpg keywords: best model for openclaw, openclaw best model, best models for openclaw, best llm for openclaw, openclaw model ranking, best ai model for openclaw, openclaw models list, openclaw supported models, best gemini model for openclaw, openclaw model recommendations, llm selection, claude, gpt, haimaker, local ai, ollama faq: - question: "What is the best AI model for OpenClaw in 2026?" answer: "For most users, Claude Sonnet 4.6 is the best default — it handles daily assistant tasks, coding, and multi-step work well without a scary bill. For coding specifically, Claude Opus 4.8 is worth the premium, especially since its price dropped to roughly a third of earlier Opus generations. For privacy-sensitive work, use open-weight models like Qwen3 Coder or Kimi K2.7 through haimaker.ai." - question: "What is the cheapest AI model for OpenClaw?" answer: "DeepSeek V4 Flash is the cheapest serious cloud option — sub-dollar per million tokens, cheap enough to run an agent around the clock for a few dollars a month. For open-weight models, haimaker.ai offers MiniMax M3, Qwen3 Coder, DeepSeek V4, and others at 5% below market rate." - question: "How do I add haimaker.ai as a provider in OpenClaw?" answer: "Add haimaker to the models.providers section in ~/.openclaw/openclaw.json with baseUrl 'https://api.haimaker.ai/v1', your API key, and api type 'openai-completions'. Then add your desired models (e.g., qwen3-coder, llama-3.3-70b). The API is fully OpenAI-compatible." - question: "Which Gemini model works best with OpenClaw?" answer: "Gemini 3 Pro is best for research and document analysis thanks to its 1M+ token context window. Gemini 3 Flash is better for high-volume, simple tasks at $0.075/$0.30 per million tokens. Both work through haimaker.ai or Google's API directly." - question: "Which OpenClaw model should I choose?" answer: "For daily work, start with Claude Sonnet 4.6. For coding, use Claude Opus 4.8. For long documents, Gemini 3.1 Pro. For budget-friendly tasks, MiniMax M3 or DeepSeek V4 Flash through haimaker.ai. Most users run 2-3 models and route between them based on task complexity." --- Everyone asks which model to use. The honest answer: it depends on what you're doing and how much you want to spend. > **Note:** Clawdbot has been rebranded to **OpenClaw** — same powerful AI agent platform, new name. Learn more at [openclaw.ai](https://openclaw.ai). OpenClaw supports a dozen providers. Anthropic, OpenAI, Google, open-source models through haimaker.ai. Each has tradeoffs around cost, capability, and where your data ends up. Here's how I think about picking one. ## OpenClaw supported models list OpenClaw can use any model exposed through a supported provider or an OpenAI-compatible endpoint. The practical supported list is wider than the default config. | Provider | Models Worth Starting With | Best Use | |---|---|---| | Haimaker | MiniMax M3, Qwen3 Coder, Kimi K2.7, DeepSeek V4 | One API key across many cheap models | | Anthropic | Claude Opus 4.8, Claude Sonnet 4.6 | Coding and hard agent tasks | | OpenAI | GPT-5.6, GPT-5.5, GPT-5.4 Mini | General work, reliable API behavior | | Google | Gemini 3 Flash, Gemini 3.1 Pro | Long context and free-tier usage | | xAI | Grok 4.5, Grok 4.1 Fast, Grok Code Fast | Cheap huge context or code-specific work | | DeepSeek | DeepSeek V4 Pro, DeepSeek V4 Flash | Low-cost coding and reasoning fallback | | Ollama | Qwen3.6, Gemma 4, gpt-oss | Local/private inference | If the model is OpenAI-compatible, the rule is simple: add the provider, add the model ID, then allowlist the fully qualified model name. That covers Haimaker, Ollama, OpenRouter, Together, private vLLM deployments, and most hosted inference gateways. ## Price, capability, privacy These three things compete with each other. You can optimize for two, maybe, but rarely all three. #### Price Token pricing varies wildly. Premium Claude models cost dozens of times more per token than budget options like Grok 4.1 Fast or DeepSeek V4 Flash — for what are, in many cases, similar outputs. One notable shift: Claude Opus 4.8 launched at roughly a third of the old Opus price, which collapsed the gap between "premium" and "mid-tier." For most assistant tasks, a mid-tier model makes sense. Claude Sonnet 4.6 gives you most of Opus's capability at a fraction of the cost. #### Capability Benchmarks lie. For OpenClaw, what actually matters: - **Tool calling** – Can it invoke shell commands and APIs without fumbling the syntax? - **Context tracking** – Does it remember what you said 50 messages ago? - **Code quality** – When it writes code, does it run? - **Speed** – How long before it starts responding? #### Privacy Cloud APIs mean your prompts hit external servers. For personal finance, health data, or proprietary code, that's a problem. You can self-host open-source models, but that requires hardware and tolerance for latency. ## Recommendations by use case #### Daily assistant work **Claude Sonnet 4.6** (mid-tier pricing) Calendar, email, research, general queries. Sonnet handles all of it without breaking the bank. Fast enough for real-time chat, smart enough for multi-step tasks. **Cheaper option: [GPT-5.4 Mini](/blog/gpt-4o-mini-openclaw)** (sub-dollar tier) Fine for simple stuff. Quality drops on anything complex, but at a fraction of the cost, sometimes that's the right call. #### Coding and automation **[Claude Opus 4.8](/blog/claude-opus-4-6-openclaw)** (premium tier, but ~3x cheaper than earlier Opus generations) When the code needs to actually work, Opus is worth the premium. It handles multi-file edits and complex debugging better than anything else I've used, and the 4.8 price cut makes it viable as a daily coding default rather than a special-occasion model. **Alternative:** Sonnet 4.6 with extended thinking enabled. Pay more per reasoning token only when you need the horsepower. #### Research and document analysis **Gemini 3.1 Pro** (mid-tier pricing) The 1M+ token context window lets you throw entire codebases at it. Good at synthesizing information across long documents. #### Gemini models in OpenClaw Google's Gemini family is worth a closer look if you're doing document-heavy work or want a solid mid-range option. **Gemini 3.1 Pro** is the workhorse. That million-token context window means you can feed it an entire repo and ask it to find the bug. For long-document analysis, contract review, or codebase Q&A, nothing else comes close on context length. **[Gemini 3 Flash](/blog/gemini-3-flash-preview-openclaw)** (deep sub-dollar pricing) is the speed option. It's cheap, fast, and surprisingly capable for simpler tasks. If you're routing high-volume queries and don't need deep reasoning, Flash handles it well. To add Gemini models through haimaker.ai, add them to your provider config: ```json5 { models: { providers: { haimaker: { // ... existing config models: [ // ... existing models { id: "google/gemini-3-pro", name: "Gemini 3 Pro" }, { id: "google/gemini-3-flash", name: "Gemini 3 Flash" } ] } } } } ``` Or use them directly through Google's API if you prefer. OpenClaw supports Google as a first-party provider. #### Running Gemma 4 locally with OpenClaw If you want zero API costs, **Gemma 4 + Ollama** is the best local model for OpenClaw right now. Google's Gemma 4 8B runs on any Mac with 16GB unified memory and handles code reading, simple edits, and boilerplate without sending anything to external servers. Install Ollama, pull the model, then point OpenClaw at it: ```bash ollama pull gemma4 ``` ```json5 { models: { providers: { ollama: { baseUrl: "http://localhost:11434/v1", api: "openai-completions", models: [{ id: "gemma4:latest", name: "Gemma 4 8B", contextWindow: 131072, maxTokens: 8192 }] } } }, agents: { defaults: { model: { primary: "ollama/gemma4:latest" } } } } ``` Full walkthrough: [How to set up Gemma 4 with OpenClaw using Ollama](/blog/gemma-4-ollama-openclaw-setup/). #### Privacy-sensitive work **Qwen3 Coder or Llama 3.3 70B through haimaker.ai** Open-source models, routed through compliant infrastructure. Your prompts stay off the big providers' training pipelines. For maximum paranoia, self-host with Ollama or vLLM. You'll need serious hardware (2x A100 or equivalent) and patience for higher latency. **Hybrid approach:** Use cloud APIs for general work, switch to open-source for sensitive tasks. Clawdbot makes this easy with model overrides. ## Provider comparison #### Anthropic (Claude) Premium pricing ($3-$75 per million output tokens). Best tool calling and instruction following. No training on API data by default. Claude has become the default for coding agents. The tool use is just more reliable than the alternatives. #### OpenAI (GPT) Mid-tier pricing ($0.60-$15). Solid general performance, fast responses. [GPT-4o](/blog/gpt-4o-openclaw) is a good all-rounder. The mini variant works well for high-volume, simple tasks. #### Google (Gemini) Competitive pricing ($1.25-$10). That massive context window is the selling point. Great for document-heavy workflows. #### Open source through haimaker.ai 5% below market rate ($0.10-$5 per million tokens). Routes requests across GPU providers for cost and latency optimization. Avoids the compliance headaches of sending data to US hyperscalers. The API is OpenAI-compatible: ```bash curl https://api.haimaker.ai/v1/chat/completions \ -H "Authorization: Bearer $HAIMAKER_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "llama-3.3-70b", "messages": [{"role": "user", "content": "Hello!"}] }' ``` Swap your base URL to `https://api.haimaker.ai/v1` and you're done. GET $10 FREE CREDITS ON HAIMAKER ## OpenClaw configuration Setting your default model in `~/.openclaw/openclaw.json`: ```json5 { agents: { defaults: { model: { primary: "anthropic/claude-sonnet-4-6" } } } } ``` Switch models mid-session with `/model opus` or `/model haimaker/llama-3.3-70b`. > **Migration note:** If you're coming from Clawdbot, your config files are automatically migrated. The CLI now uses `openclaw` instead of `clawdbot` commands. #### Adding haimaker.ai as a provider in OpenClaw ```json5 { env: { HAIMAKER_API_KEY: "sk-..." }, agents: { defaults: { model: { primary: "haimaker/llama-3.3-70b" } } }, models: { mode: "merge", providers: { haimaker: { baseUrl: "https://api.haimaker.ai/v1", apiKey: "${HAIMAKER_API_KEY}", api: "openai-completions", models: [ { id: "llama-3.3-70b", name: "Llama 3.3 70B" }, { id: "qwen3-coder", name: "Qwen3 Coder" }, { id: "mistral-large", name: "Mistral Large" } ] } } } } ``` Prefer to skip the hand-editing? `npx -y @haimaker/connect --openclaw` writes that exact provider block into `~/.openclaw/openclaw.json` for you; the [connect guide](/connect) covers the other agents it handles too. ## Which model should I choose? If you're still unsure, here's the quick decision tree: **"I just want something that works"** — Claude Sonnet 4.6. It handles 80% of tasks well and the pricing is reasonable. Start here. **"I'm writing production code"** — Claude Opus 4.8. The extra cost pays for itself when you're debugging a gnarly async issue at 2am and the model actually gets it right the first time. **"I need to process long documents"** — Gemini 3.1 Pro. Nothing else gives you a million tokens of context. Feed it the whole repo, the whole contract, the whole thread. **"I need it free"** — [Gemma 4 with Ollama](/blog/gemma-4-ollama-openclaw-setup/) is the best free local model for OpenClaw right now. Runs on any 16GB Mac. Gemini Flash also has a free tier. See our [free models guide](/blog/best-free-models-for-openclaw). **"I need it cheap"** — MiniMax M3 or DeepSeek V4 Flash through haimaker.ai for simple tasks, GLM-5.2 when you need more reasoning. See our [cost comparison guide](/blog/cheapest-models-openclaws). **"I want my data to stay private"** — Qwen3 Coder or Llama 3.3 70B through haimaker.ai. Or [self-host with Ollama](/blog/best-local-models-for-openclaw) if you have the hardware. **"I want to use Gemini"** — Gemini 3.1 Pro for quality, Gemini 3 Flash for speed and cost. Both available through haimaker.ai or Google's API directly. The honest answer is that most people should run two or three models and route between them. Use a cheap model for simple tasks, a mid-tier model for daily work, and a premium model for the hard stuff. Haimaker's routing engine makes this easy to set up. ## OpenClaw models list Full pricing, setup, and use-case breakdown for every model supported in OpenClaw: **Anthropic:** [Claude Opus 4.6](/blog/claude-opus-4-6-openclaw) · [Opus 4.5](/blog/claude-opus-4-5-openclaw) · [Opus 4.1](/blog/claude-opus-4-1-openclaw) · [Sonnet 4.6](/blog/claude-sonnet-4-6-openclaw) · [Sonnet 4.5](/blog/claude-sonnet-4-5-openclaw) · [Haiku 4.5](/blog/claude-haiku-4-5-openclaw) **OpenAI:** [GPT-5.3](/blog/gpt-5-3-chat-openclaw) · [GPT-5 Codex](/blog/gpt-5-codex-openclaw) · [GPT-5 Pro](/blog/gpt-5-pro-openclaw) · [GPT-5](/blog/gpt-5-openclaw) · [GPT-5 Mini](/blog/gpt-5-mini-openclaw) · [GPT-5 Nano](/blog/gpt-5-nano-openclaw) · [GPT-5.4 Pro](/blog/gpt-5.4-pro-openclaw) · [GPT-5.4](/blog/gpt-5.4-openclaw) · [GPT-5.2](/blog/gpt-5.2-openclaw) · [GPT-5.1](/blog/gpt-5.1-openclaw) · [GPT-4.1](/blog/gpt-4.1-openclaw) · [GPT-4.1 Mini](/blog/gpt-4.1-mini-openclaw) · [GPT-4.1 Nano](/blog/gpt-4.1-nano-openclaw) · [GPT-4o](/blog/gpt-4o-openclaw) · [GPT-4o Mini](/blog/gpt-4o-mini-openclaw) · [GPT-4 Turbo](/blog/gpt-4-turbo-openclaw) · [GPT-4](/blog/gpt-4-openclaw) · [o3](/blog/o3-openclaw) · [o3 Mini](/blog/o3-mini-openclaw) · [o4 Mini](/blog/o4-mini-openclaw) · [o1](/blog/o1-openclaw) · [o1 Mini](/blog/o1-mini-openclaw) **Google:** [Gemini 3.1 Pro](/blog/gemini-3.1-pro-preview-openclaw) · [Gemini 3 Flash](/blog/gemini-3-flash-preview-openclaw) · [Gemini 2.5 Pro](/blog/gemini-2.5-pro-openclaw) · [Gemini 2.5 Flash](/blog/gemini-2.5-flash-openclaw) · [Gemini 2.0 Flash](/blog/gemini-2.0-flash-001-openclaw) **xAI:** [Grok 4.20](/blog/grok-4-20-beta-openclaw) · [Grok 4](/blog/grok-4-openclaw) · [Grok 4 Fast](/blog/grok-4-fast-openclaw) · [Grok 4.1 Fast](/blog/grok-4-1-fast-openclaw) · [Grok Code Fast](/blog/grok-code-fast-openclaw) · [Grok 3](/blog/grok-3-openclaw) · [Grok 3 Mini](/blog/grok-3-mini-openclaw) · [Grok 3 Mini Fast](/blog/grok-3-mini-fast-openclaw) · [Grok 2](/blog/grok-2-openclaw) · [Grok 2 Vision](/blog/grok-2-vision-openclaw) **DeepSeek:** [DeepSeek R1](/blog/deepseek-r1-openclaw) · [DeepSeek V3.2](/blog/deepseek-v3.2-openclaw) · [DeepSeek V3.1](/blog/deepseek-chat-v3.1-openclaw) · [DeepSeek V3](/blog/deepseek-chat-openclaw) **MiniMax:** [M2.5](/blog/minimax-m2.5-openclaw) · [M2.5 Lightning](/blog/minimax-m2.5-lightning-openclaw) · [M2.1](/blog/minimax-m2.1-openclaw) · [M2.1 Lightning](/blog/minimax-m2.1-lightning-openclaw) · [M2](/blog/minimax-m2-openclaw) **Zhipu:** [GLM-5](/blog/glm-5-openclaw) · [GLM-4.7](/blog/glm-4.7-openclaw) · [GLM-4.7 Flash](/blog/glm-4.7-flash-openclaw) · [GLM-4.6](/blog/glm-4.6-openclaw) · [GLM-4.6 Exacto](/blog/glm-4.6-exacto-openclaw) **Meta:** [Llama 4 Maverick](/blog/llama-4-maverick-openclaw) · [Llama 4 Scout](/blog/llama-4-scout-openclaw) **More:** [Qwen3.5 397B](/blog/qwen3-5-397b-a17b-openclaw) · [Qwen3 Max](/blog/qwen3-max-openclaw) · [Qwen3 Coder](/blog/qwen3-coder-openclaw) · [Qwen3 Coder Plus](/blog/qwen3-coder-plus-openclaw) · [Qwen 2.5 Coder 32B](/blog/qwen-2.5-coder-32b-instruct-openclaw) · [Kimi K2](/blog/kimi-k2-thinking-openclaw) · [Kimi K2.5](/blog/kimi-k2.5-openclaw) · [MiMo V2 Flash](/blog/mimo-v2-flash-openclaw) · [UI-TARS 1.5](/blog/ui-tars-1.5-7b-openclaw) ## Bottom line There's no best model. There's the right model for what you're doing. - **Cheap:** MiniMax M3, DeepSeek V4 Flash, or open-source through haimaker.ai - **Capable:** Opus 4.8 or Gemini 3.1 Pro - **Private:** Open-source through haimaker.ai or self-hosted Most people should start with Claude Sonnet 4.6. It handles most tasks well and won't run up a scary bill. Adjust from there based on what you actually need. EXPLORE HAIMAKER --- *Ready to set up your own OpenClaw agent? Visit [openclaw.ai](https://openclaw.ai) to get started or check out the [OpenClaw documentation](https://docs.openclaw.ai) for detailed configuration options.*